Fix LT-22630: Custom first vernacular fields defaulting to analysis

[jasonleenaylor]

Quick Summary

Fixes LT-22630: since FW 9.3.7, typing in a custom single-string field configured for First Vernacular came out in the first analysis writing system.

The LT-22145 fix overrode StringSliceView.WsPending to always return DefaultAnalWs and swallow writes, turning the one-shot pending-ws mechanism into a permanent forced writing system for every keystroke in these slices. This PR scopes the override to the case LT-22145 was actually about — an empty field, where there is no adjacent text to inherit a ws from — and picks the ws the slice was configured with: the constructor-specified ws (custom fields), else the wsempty config default, else DefaultAnalWs. When the field has content, the getter returns -1 so typing inherits the surrounding text's writing system.

• StringSlice.cs: empty-field-scoped WsPending getter; DefaultWs (the wsempty attribute) now stored on the view; empty-check handles String and MultiString/MultiUnicode props.
• StringSliceUtils.cs: the ws fallback chain as a pure GetWsForEmptyField function.
• StringSliceUtilsTests.cs: unit tests for the chain, including that negative magic-ws constants are never treated as real handles.

CI-ready checklist

• Commit messages follow .github/commit-guidelines.md (subject ≤ 72 chars, no trailing punctuation; if body present, blank line then ≤ 80-char lines).
• No whitespace warnings locally:

  git fetch origin
  git log --check --pretty=format:"---% h% s" origin/<base>..
  git diff --check --cached

• Builds/tests pass locally (or I've run the CI-style build via Bash script or MSBuild).
• If this is core-developer AI-assisted work, I followed Docs/workflows/ai-pr-workflow.md and ran pr-preflight or the equivalent branch-readiness review before requesting review.
• For any Src/** folders touched, corresponding AGENTS.md files are updated or explicitly confirmed still accurate.

Notes for reviewers (optional)

• Behavior for LT-22145's fields (Source on Sense, Reference on Example) is unchanged when empty: they still default to first analysis. The change additionally un-breaks typing inside existing non-analysis text in plain string fields, which the old override forced to analysis per keystroke.
• The only shipped wsempty user (Data Notebook Title, wsempty="analysis") resolves to DefaultAnalWs, identical to the last-resort fallback, so honoring it changes nothing for existing configurations.
• DetailControlsTests pass locally (67 passed, 1 skipped, including 4 new tests). Manual acceptance tests for both tickets in live FLEx are still pending; this is a draft until that verification is done.

🤖 Generated with Claude Code

---

This change is 

[github-actions]

NUnit Tests

    1 files  ±0      1 suites  ±0   9m 47s ⏱️ - 1m 2s
4 306 tests +4  4 233 ✅ +5  73 💤 ±0  0 ❌  - 1 
4 315 runs  +4  4 242 ✅ +5  73 💤 ±0  0 ❌  - 1 

Results for commit d6a977f. ± Comparison against base commit b30d631.