Document Sentry fingerprinting for timeout grouping

[nertzy]

Because rack-timeout raises at whatever line is executing when the request clock expires, similar timeouts get distinct stack traces and Sentry groups them as many separate issues — the same problem the existing Rollbar guide addresses.

This adds doc/sentry.md (mirroring doc/rollbar.md) showing a before_send hook that:

• groups timeouts by transaction (controller#action) via a custom fingerprint instead of by stack trace,
• leads the issue title with the route while preserving the original wait/timeout message,
• matches RequestTimeoutException and the Rack::Timeout::Error base (so RequestTimeoutError and RequestExpiryError are covered regardless of where Sentry's capture middleware sits in the stack),
• leaves transaction-less timeouts (raised before controller dispatch) on Sentry's default grouping.

Also links the new page from the README's doc list, alongside the Rollbar entry.

[nertzy]

(I used AI to build this but it was human-reviewed as well)