release-25.3: pkg/util/log: Implement handling for oversized log messages in bufferedSink

[blathers-crl]

Backport 1/1 commits from #157964 on behalf of @Abhinav1299.

---

Previously, when a single log message exceeded the configured max-buffer-size
for a buffered sink with exit-on-error enabled, the error would propagate up
and trigger process termination. This was overly aggressive for what amounts
to a logging configuration issue - a single oversized SQL query (e.g., with a
multi-megabyte string literal) could crash an entire CockroachDB node.

This commit modifies bufferedSink.output() to detect the errMsgTooLarge error
and handle it gracefully. When an oversized message is encountered, instead of
propagating the error, we drop the message and log a warning via Ops.Warningf()
indicating that the message exceeded the buffer size limit. This allows the
node to continue operating normally while still providing visibility into the
issue through logged warnings.

The implementation uses a two-phase approach to avoid deadlock: first, while
holding the sink's mutex, we detect the oversized message and set a flag with
the relevant information; then, after releasing the lock, we emit the warning.
This is necessary because calling Ops.Warningf() while holding the mutex would
cause the warning message to attempt re-entry into the same sink, resulting in
a deadlock when it tries to acquire the already-held lock.

This resolves #152635

Part of: CRDB-53951
Epic: CRDB-56325
Release note: None

---

Release justification: This PR fixes the CRDB process termination when a log message greater than max-buffer-size of a sink is encountered when exit-on-error flag is enabled.

[blathers-crl]

Thanks for opening a backport.

Before merging, please confirm that the change does not break backwards compatibility and otherwise complies with the backport policy. Include a brief release justification in the PR description explaining why the backport is appropriate. All backports must be reviewed by the TL for the owning area. While the stricter LTS policy does not yet apply, please exercise judgment and consider gating non-critical changes behind a disabled-by-default feature flag when appropriate.

[cockroach-teamcity]

This change is 

[dhartunian]

Release justification: this is a low risk improvement to our error behavior for critical log sinks that can cause unnecessary node crashes.