Fix unused_import for re-exported modules

[catlover-bot]

Summary

Fix unused_import incorrectly removing an umbrella import when symbols are
used through one of that module's @_exported imports.

SourceKit reports the referenced symbol's defining module, not the umbrella
module imported by the source file. The rule now resolves exported imports
with swift-api-digester before classifying an umbrella import as unused.

Implementation

• Detect API-digester Import nodes marked Exported.
• Preserve allowed_transitive_imports and require_explicit_imports.
• Cache successful lookups by module and compiler context.
• Retry transient lookup failures.
• Forward only module-loading compiler arguments and omit plugin paths.
• Fall back to the previous behavior when the digester cannot run.

Regression coverage

• Remove the manual Foundation -> CoreFoundation workaround from the
  existing CGFloat example.
• Test exported versus ordinary imports, argument filtering, cache reuse,
  failure retry, and separate compiler contexts.
• Verify a real A -> @_exported import B package and a plain-import control.

Validation

• Resolver unit tests: 4 passed
• UnusedImportRuleGeneratedTests: passed
• SwiftPM: 1,091 tests in 375 suites passed
• SwiftLint self-lint: 0 violations in 713 files
• Bazel: 19/19 test targets passed
• Universal release build: passed
• Exported-import reproduction: 0 violations
• Plain unused-import control: 1 expected violation

Opened as a draft for cross-platform CI validation of the API-digester
integration.

Closes #5528

[SwiftLintBot]

	1 Warning
⚠️	Big PR

	20 Messages
📖	Building this branch resulted in a binary size of 28444.16 KiB vs 28415.46 KiB when built on main (0% larger).
📖	Linting Aerial with this PR took 0.68 s vs 0.69 s on main (1% faster).
📖	Linting Alamofire with this PR took 0.97 s vs 0.98 s on main (1% faster).
📖	Linting Brave with this PR took 6.26 s vs 6.26 s on main (0% slower).
📖	Linting Brigade with this PR took 18.12 s vs 18.21 s on main (0% faster).
📖	Linting DuckDuckGo with this PR took 29.99 s vs 29.85 s on main (0% slower).
📖	Linting Firefox with this PR took 11.38 s vs 11.4 s on main (0% faster).
📖	Linting Kickstarter with this PR took 7.32 s vs 7.34 s on main (0% faster).
📖	Linting Moya with this PR took 0.37 s vs 0.37 s on main (0% slower).
📖	Linting NetNewsWire with this PR took 2.49 s vs 2.52 s on main (1% faster).
📖	Linting Nimble with this PR took 0.57 s vs 0.56 s on main (1% slower).
📖	Linting PocketCasts with this PR took 7.38 s vs 7.41 s on main (0% faster).
📖	Linting Quick with this PR took 0.36 s vs 0.36 s on main (0% slower).
📖	Linting Realm with this PR took 2.98 s vs 2.96 s on main (0% slower).
📖	Linting Sourcery with this PR took 1.63 s vs 1.67 s on main (2% faster).
📖	Linting Swift with this PR took 4.54 s vs 4.52 s on main (0% slower).
📖	Linting SwiftLintPerformanceTests with this PR took 0.18 s vs 0.18 s on main (0% slower).
📖	Linting VLC with this PR took 1.31 s vs 1.3 s on main (0% slower).
📖	Linting Wire with this PR took 16.34 s vs 16.29 s on main (0% slower).
📖	Linting WordPress with this PR took 10.44 s vs 10.51 s on main (0% faster).

Generated by 🚫 Danger