Optimize search snippet and guard line processing

[Widthdom]

Summary

• Stream search snippet line scanning so bounded excerpts no longer split or normalize whole content arrays.
• Cap tracked snippet match indexes while preserving dropped-match accounting.
• Precompute guarded-search primary terms, bound guard window extraction, cache repeated guard windows, and reuse prepared match-line context for enclosing-symbol lookup.

Validation

• dotnet build
• dotnet test tests/CodeIndex.Tests/CodeIndex.Tests.csproj --filter FullyQualifiedName~SearchSnippetFormatterTests
• dotnet test tests/CodeIndex.Tests/CodeIndex.Tests.csproj --filter FullyQualifiedName~DbSearchReaderIssueTests
• dotnet test tests/CodeIndex.Tests/CodeIndex.Tests.csproj --filter FullyQualifiedName~Search_EnclosingSymbolUsesPreparedMatchLineContext_Issue3086
• dotnet test tests/CodeIndex.Tests/CodeIndex.Tests.csproj
• dotnet run --project tools/CodeIndex.Changelog -- check
• dotnet ./src/CodeIndex/bin/Debug/net8.0/cdidx.dll status --check --json
• Codex adversarial review: No blocking/actionable issues found.

Documentation / Changelog

• Added bilingual changelog fragments:
  • changelog.d/unreleased/3083.fixed.md
  • changelog.d/unreleased/3084.fixed.md
  • changelog.d/unreleased/3085.fixed.md
  • changelog.d/unreleased/3086.fixed.md
  • changelog.d/unreleased/3087.fixed.md
  • changelog.d/unreleased/3088.fixed.md

Follow-up Candidates

• None.

Fixes #3083
Fixes #3084
Fixes #3085
Fixes #3086
Fixes #3087
Fixes #3088