Instrument src/github/installation_token.rs with tracing

[coderabbitai]

Summary

Add observability to the installation-token acquisition code path in src/github/installation_token.rs using the tracing crate, which is already a project dependency.

Motivation

The installation-token acquisition code lacks logging at critical operational boundaries. Without instrumentation, diagnosing production failures (e.g. buffer rejections, API errors) requires code-level investigation rather than log inspection.

Proposed change

• Add #[instrument] to installation_token_with_buffer and installation_token_with_factory (and installation_token_with_client_and_time if appropriate).
• Emit tracing::info! on successful token acquisition (including expiry timestamp, excluding token secret).
• Emit tracing::warn! when a token is rejected due to the expiry buffer.
• Emit tracing::error! on acquisition failure, including the mapped GitHubError variant.
• Document the expected operational signals (log levels, fields) in docs/developers-guide.md.

Acceptance criteria

• All async entry points carry #[instrument] (with skip for sensitive fields as appropriate).
• At minimum, info/warn/error events are emitted at the three points listed above.
• Developer documentation is updated.
• make check-fmt, make lint, and make test all pass.

Background

Identified during review of PR #77 (requested by @leynos).