fix(lsp): await background init task on shutdown

[ebbsanchez]

Description

Fixes the shutdown path introduced by non-blocking LSP startup by keeping the background initialization task handle and awaiting it with a bounded timeout after cancellation is sent.

This keeps serve_with non-blocking during startup, but makes shutdown observe the background task instead of dropping the JoinHandle. If the task panics, it is now logged; if it does not stop promptly, shutdown logs a timeout and continues.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update
• Refactoring (no functional changes)

Related Issues

Fixes #196

Checklist

• I have read the CONTRIBUTING guide
• My code follows the project's coding style
• I have added tests that prove my fix/feature works
• All new and existing tests pass
• I have updated the documentation accordingly
• I have updated the CHANGELOG.md (for user-facing changes)

Verification

• cargo +nightly fmt --check
• cargo test -p mcpls-core graceful_degradation_tests -- --nocapture
• cargo clippy -p mcpls-core --all-targets --all-features -- -D warnings

Additional Notes

I did not update CHANGELOG.md because this is an internal shutdown/diagnostic behavior fix without a user-facing CLI/API change.

[ebbsanchez]

CI note: I checked the current failures. The CodeQL jobs fail before analyzing this patch because the pull_request_target workflow's actions/checkout@v7 step refuses to checkout fork PR code with allow-unsafe-pr-checkout: false. The regular CI run is marked action_required, which looks like the usual maintainer approval gate for fork PRs.

Local verification for the patch is still:

• cargo +nightly fmt --check
• cargo test -p mcpls-core graceful_degradation_tests -- --nocapture
• cargo clippy -p mcpls-core --all-targets --all-features -- -D warnings