Backport(v1.19): fix(buffer): decrement queue_size in ensure so failed purge does not leak (#5475)

[github-actions]

Which issue(s) this PR fixes:
Backport #5475
Fixes #5468

What this PR does / why we need it:
In Fluent::Plugin::Buffer#purge_chunk, @queue_size_metrics is only decremented inside begin/rescue. When chunk.purge raises (e.g. unlink/close failure on the buffer path), the error is logged and swallowed but the sub is skipped. The chunk is already removed from @dequeued, so it is never retried — the queued byte counter ratchets toward total_limit_size, making storable? permanently false and causing a spurious BufferOverflowError on a near-empty buffer.

This moves bytesize = chunk.bytesize above the begin and moves @queue_size_metrics.sub(bytesize) into an ensure block so the counter is always released exactly once per purge_chunk call, regardless of whether purge succeeds.

Why safe:

• ensure runs exactly once per call → no double-decrement.
• Every chunk reaching purge_chunk was previously enqueued (dequeue/emit path), so the sub is always balanced.
• If chunk.purge raises, the chunk is lost on filesystem error (already a logged, separate failure); this change only preserves counter integrity so storable? no longer spuriously goes false.

Docs Changes:
None (behavior fix; inline comment added).

Release Note:

• buffer: fix spurious BufferOverflowError caused by queue_size leaking when a chunk purge fails