This PR massively reorganizes the Test dialect's source files. It moves
manually-written op hooks into `TestOpDefs.cpp`, moves format custom
directive parsers and printers into `TestFormatUtils`, adds missing
comment blocks, and moves around where generated source files are
included for types, attributes, enums, etc. into their own source file.

This will hopefully help navigate the test dialect source code, but also
speeds up compile time of the test dialect by putting generated source
files into separate compilation units.

This also sets up the test dialect to shard its op definitions, done in
the next PR.

stack-info: PR: https://github.com/llvm/llvm-project/pull/89424, branch: users/mogball/pr_2

Adds an option to `mlir-tblgen -gen-op-defs` `op-shard-count=N` that divides the
op class definitions and op list into N segments, e.g.

```
// mlir-tblgen -gen-op-defs -op-shard-count=2

void FooDialect::initialize() {
  addOperations<
  >();
  addOperations<
  >();
}

```

When split across multiple source files, this can help significantly improve
dialect compile time for dialects with a large opset.

stack-info: PR: https://github.com/llvm/llvm-project/pull/89423, branch: users/mogball/pr_1

Removes an unused static function `IsOverloaded` from `SemaOverload.cpp`
that is unused after #88731.

Upcoming patches will add TP relative stack checks for Android, and
Linux currently uses the default GOT based stack protector.

Use the datalayout directly to determine the correct `cvta` instruction
for converting shared/local/const pointers. This is cleaner as it
eliminates the need to keep a redundant copy of this info in the TM and
makes clear which address spaces short pointers are applicable for.

Bit-fields of bit-precise integer type do not promote to int, but
instead promote to the type of the field.

Fixes #87641

num_gangs takes an 'int-expr-list', for 'parallel', and an 'int-expr'
for 'kernels'. This patch changes the parsing to always parse it as an
'int-expr-list', then correct the expression count during Sema. It also
implements the rest of the semantic analysis changes for this clause.

Reapply "[Clang][Sema] Fix crash when 'this' is used in a dependent class scope function template specialization that instantiates to a static member function (#87541, #88311)" (#88731)

Reapplies #87541 and #88311 (again) addressing the bug which caused
expressions naming overload sets to be incorrectly rebuilt, as well as
the bug which caused base class members to always be treated as overload
sets.

The primary change since #88311 is `UnresolvedLookupExpr::Create` is called directly in `BuildPossibleImplicitMemberExpr` with `KnownDependent` as `true` (which causes the expression type to be set to `ASTContext::DependentTy`). This ensures that any further semantic analysis involving the type of the potentially implicit class member access expression is deferred until instantiation. 

Need to relax assertion and check ReuseShuffleIndices is not empty, if
the root phi node has reorder indices.

The compiler should not take into account the type of the cmp
instruction, otherwise it may treat the size incorrectly and it may lead
to incorrect codegen.

The implementation follows the ELF implementation.

[VectorCombine] foldShuffleOfShuffles - fold "shuffle (shuffle x, undef), (shuffle y, undef)" -> "shuffle x, y" (#88743)

Another step towards cleaning up shuffles that have been split, often across bitcasts between SSE intrinsic.

Strip shuffles entirely if we fold to an identity shuffle.

Extra tests with strides with different signs for
https://github.com/llvm/llvm-project/pull/88039.

The argument types are not modeled as children of TypeTraitExpr,
therefore they are not dumped with the default implementation.

Dumping them is really useful for ad-hoc debugging, context #89358

This removes `m_forward_decl_die_to_compiler_type` which is a map from
`const DWARFDebugInfoEntry *` to `lldb::opaque_compiler_type_t`. This
map is currently used in `DWARFASTParserClang::ParseEnum` and
`DWARFASTParserClang::ParseStructureLikeDIE` to avoid creating duplicate
CompilerType for the specific DIE. But before entering these two
functions in `DWARFASTParserClang::ParseTypeFromDWARF`, we already
checked with `SymbolFileDWARF::GetDIEToType()` if we have a Type created
from this DIE to avoid trying to parse the same DIE twice. So, this map
is unnecessary and not useful.

#84050 resolves class member access expressions naming members of the
current instantiation prior to instantiation. In testing, it has
revealed a mem-initializer in the move constructor of
`invocable_with_telemetry` that uses an unparenthesized comma expression
to initialize a non-static data member of pointer type. This patch fixes it.

Need to check all possible entries, before trying looking for the
minbitwidth in the user node. Otherwise we may incorrectly get
signedness info.

The assertion doesn't work if there are multiple declarations for
a variable involved.

Move the iterator declarations into the if statements and return
std::nullopt explicitly.

Some docs were emitted into the wrong location (Polynomial/ instead of
Dialect/). Furthermore, `-gen-dialect-docs` subsumes
`-gen-attr/typedef-docs` so the latter are not required.

Add a top-level entry that includes both other files in a proper order.

COMMON block names must be declared in the same scoping unit in
which the OpenMP directive or clause appears, but OpenMP
constructs must not be considered as scoping units. Instead,
consider only program units and block constructs as such.

As suggested in https://github.com/llvm/llvm-project/pull/88039, add
extra documentation for reasoning in isDependent.

PR: https://github.com/llvm/llvm-project/pull/89381

Failed on main trunk: https://godbolt.org/z/edWMz8chE

This transformation, inspired by what is done in
hoist_redundant_transfers, hoists pairs of extract/broadcast operations
out of scf.for loops.

It changes a loop of the form:

```
%res = scf.for _ = _ to _ step _ iter_args(%iarg = %v) -> (t1) {
  %e = vector.extract %iarg : t1 to t2
  %u = "some_use"(%e) : (t2) -> t2
  %b = vector.broadcast %u : t2 to t1
  scf.yield %b : t1
}
```

into the following:

```
%e = vector.extract %v: t1 to t2
%res' = scf.for _ = _ to _ step _ iter_args(%iarg = %e) -> (t2) {
  %u' = "some_use"(%iarg) : (t2) -> t2
  scf.yield %u' : t2
}
%res = vector.broadcast %res' : t2 to t1
```

The modulemap file is not generated anymore, so it's just part of our
list of includes and gets installed like every other header. We don't
need a special step to install it anymore. This was overlooked when I
removed the generation of the modulemap file.

We can instead generate it on-the-fly when we install the headers. This
reduces the amount of boilerplate we have to re-generate whenever we
add, remove or relocate header files.

Fixes #88529

Verify-uselistorder wants to take some input IR and verify that the
uselist order is stable after roundtripping to bitcode and assembly.
This is disrupted if the file is converted between the new and old debug
info formats after parsing - while there's no functional difference, the
change to the in-memory representation of the IR modifies the uselist.
This patch changes verify-uselistorder to not convert input files
between debug info formats by default, preventing changes from being
made to the file being checked. In addition, this patch makes it so that
when we _do_ print IR in the new debug info format to bitcode or
assembly, we delete any lingering debug intrinsic declarations, ensuring
that we don't write uselist entries for them.

Closes #89192.

The original comments mention `addPreLinkLTODefaultPipeline`, but I
could not find any functions with this name, even in
https://reviews.llvm.org/D33540
(8b3be4e5) where this comment was added.
I assume they meant to refer to `buildThinLTOPreLinkDefaultPipeline` and
`buildLTOPreLinkDefaultPipeline` and so this patch uses them.

This had some repeated and overlapping conditions, which
made it more difficult to handle the new metadata scheme. Reflow
the function to handle the easy LDS cases first. For the flat/global
cases, write in a positive-enabled style where everything unhandled
hits a default cmpxchg.

Depends #89468

When exhaustive unary/binary tests fail, print the name of the function
being tested as well as the values of the inputs and outputs.

Example of a simulated failure in testing "udiv exact":

    unittests/Support/KnownBitsTest.cpp:99: Failure
Value of: checkResult(Name, Exact, Computed, {Known1, Known2},
CheckOptimality)
Actual: false (udiv exact: Inputs = ???1, ????, Computed = ???1, Exact =
0???)
    Expected: true