auto: increment master version to 17.0.189

chore(android,windows): Update crowdin strings for Kannada

auto: increment master version to 17.0.188

fix: path for ESM increment-version, and exit code

* Fixes the path for the increment-version module
* Uses exit code 50 instead of 1 to indicate 'success, but no change',
  so that low-value error exit codes as used by node.js don't
  indavertently continue the build

chore: clean up a few minor discrepancies in builder.inc.sh

epic(developer): refactor package-metadata for 17.0 🎺

chore: merge master into package-metadata

chore(developer): add --for-publishing to kmc 🎺

chore(developer): Update kmc-package xml2js dependency 🎺

chore(common): stop using xml2js.processors.firstCharLowerCase 🎺

fix(developer): non-minimal BCP 47 tags in kps should be a hint 🎺

Merge pull request #9711 from keymanapp/refactor/developer/9708-keyboard_info-example-keys-as-string

refactor(developer): .keyboard_info examples should use keys as a string 🎺

feat(developer): introduce WelcomeFile property to packages 🎺

chore: allowJs for imported schemas 🎺

chore: merge master into package-metadata 🎺

chore(linux): Improve Sentry reports

- Some uses of `KEYMAN_ROOT` (not cross-repo compatible)
- `set -eu` and `SHLVL` now set there as well

fix(linux): Explicitly initialize GTK

chore(developer): fixup signcode paths for server

Two conflicting merges meant that paths diverged. Fixes that.

chore: rename km_kbp_ to km_core_

Merge pull request #9723 from keymanapp/chore/core/9721-rename-keyboardprocessor.h-to-keyman_core_api.h

chore(core): rename keyboardprocessor.h to keyman_core_api.h

fix(common): fix schema fixer 🙀

- keyboardTest3 didn't get a fix that had applied to keyboard[3]
- regenerate using the exact same CLDR

Fixes: #9683

This change adds additional details to Sentry events so that the "All
Events" tab on keyman.sentry.io will show helpful values in the
"device" and "os" columns instead of emptry strings.

The `hash` method that we previously used calculates the hash based on
the memory location, i.e. with each program run we get a different value
for the same user. This change now uses the md5 algorithm so that the
same user always produces the same hash.