Parse quantity format suffixes invariantly

[angularsen]

Extracted from #1544 because format-string parsing is unrelated to QuantityValue and is a standalone correctness improvement for the existing formatter. Moving it out keeps the feature PR focused and lets this behavior ship independently.

Changes:

• Parse significant-digit, abbreviation-index, currency, and percent suffixes with invariant culture.
• Apply the same parsing rules to the span-based and .NET Framework-compatible code paths.

Tests:

• Add Format_CustomSpecifierSuffix_ParsesUsingInvariantCulture for abbreviation and significant-digit formats.

[claude]

Review

Small, well-scoped bug fix — no generated code, unit definitions, or public API surface touched.

Breaking changes: None. No public signatures changed. The only behavior change is that format-string suffixes (S1, A0, C, P) are now parsed with InvariantCulture instead of implicitly using CurrentCulture. This can only change behavior for callers running under a culture with a non-default PositiveSign/NegativeSign/digit set — and in those cases the old behavior was arguably already broken (a format string like "S+1" would silently fail to parse and fall through to the standard-numeric-format branch, likely throwing FormatException on "S+1" itself). Worth a one-line mention in the release notes since it's technically a behavior change, even though it's a correctness fix.

Style/conventions: Consistent with the surrounding code — mirrors the existing #if NET / #else split and doesn't introduce new abstractions. Good use of NumberStyles.Integer explicitly on the net48 path to match the default behavior of the span-based overload used on NET.

Generated code / new quantities: N/A — no changes to Common/UnitDefinitions/ or GeneratedCode/.

Correctness: Confirmed the fix is complete — grepped the codebase and QuantityFormatter.cs is the only place with this TryParse(format...) pattern, so there's no other culture-sensitive format-suffix parsing left unfixed elsewhere (e.g. UnitParser/QuantityParser).

Test coverage: Format_CustomSpecifierSuffix_ParsesUsingInvariantCulture is a solid regression test — using a cloned invariant culture with PositiveSign = "!" is a clever way to prove the fix without relying on a specific real-world culture (e.g. some cultures use "+"-like Unicode variants). One gap: the C/P (currsimilarency/percent) suffix branches aren't covered by a test, though the risk there is low since they only gate a TryParse call before throwing FormatException either way.

Nit: The PR body says "Verify both net10.0 and net48 QuantityFormatterTests (138 tests per target)" — might be worth double-checking that count is accurate/intentional if it's meant as a checklist item, but this is a documentation-only concern and doesn't affect the code.

Overall: correct, minimal, well-tested fix. LGTM.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 96%. Comparing base (7e8da49) to head (5e884c8).

Additional details and impacted files

@@          Coverage Diff           @@
##           master   #1689   +/-   ##
======================================
  Coverage      96%     96%           
======================================
  Files         456     456           
  Lines       30107   30107           
======================================
  Hits        29043   29043           
  Misses       1064    1064           

Flag	Coverage Δ
net48	96% <100%> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.