Fix Windows device paths and UTF-16 NUL detection

[Widthdom]

Summary

• Skip Windows reserved device paths and names before probing or reading file contents.
• Detect BOM-less UTF-16 LE/BE text patterns so NUL bytes from encoding do not cause binary skips or validation noise.
• Add focused FileIndexer coverage and bilingual changelog fragments.

Validation

• dotnet test tests/CodeIndex.Tests/CodeIndex.Tests.csproj --filter "FullyQualifiedName~FileIndexerTests" -p:UseSharedCompilation=false
• dotnet build CodeIndex.sln -c Release -p:UseSharedCompilation=false
• dotnet test CodeIndex.sln -c Release -p:UseSharedCompilation=false
• dotnet ./src/CodeIndex/bin/Debug/net8.0/cdidx.dll status --check --json
• Adversarial review: No blocking/actionable issues found.

Documentation / Changelog

• changelog.d/unreleased/2043.fixed.md
• changelog.d/unreleased/1829.fixed.md

Follow-up Candidates

• None.

Fixes #2043
Fixes #1829