This reverts commit 5a799219.

Use port randomization for VCS and OpenOCD

This took a lot of refactoring to make it look reasonable.
There isn't actually any functional OpenOCD test yet. But a dummy test
runs a command (and fails).

Requires OpenOCD change 4eba841bfee4b6c347423ac28851ab7ee7e75532

M-mode writes to triggers with dmode set are now ignored instead of
raising an exception.

Also added -f/--fail-fast option to gdbserver.

If the last test leaves some triggers set they should be cleaned up.

Fixes issue #30.

No longer use unittest. Now tests can return not_applicable if eg. a
desired execution mode isn't implemented on a target. Also we do a
better job killing spike processes when a test fails.

Did a lot of code cleanup, partly by using pylint.

Fix the Makefile so that if the test fails, 'make' actually fails too.

Doesn't work with Python 2.7.9 which is the default in latest Debian.

They were almost identical, so I made them actually identical.  This
will reduce the burden of writing further tests that span base ISAs.

Tests can still be specialized for XLEN with ifdefs on e.g. __riscv64.

Closes #12.

This is a pattern GCC should pick up for targets where AMOADD is faster
than LW/ADD/SW.