[fix][ml]Revert a behavior change of releasing idle offloaded ledger handle: only release idle BlobStoreBackedReadHandle

[poorbarcode]

Motivation

#24381 changed the behavior below:

• original: Pulsar only releases the offloaded ledger handle that is typed BlobStoreBackedReadHandle
• after [improve][ml]Release idle offloaded read handle only the ref count is 0 #24381, Pulsar will release the offloaded ledger handle immediately if its type is not BlobStoreBackedReadHandle, because the method lastAccessTimestamp() always returns -1 if its type is not BlobStoreBackedReadHandle

Modifications

• Revert the behavior change.
• Modify lastAccessTimestamp first, and modify pendingReading second, to avoid getting a new pendingReading and an older lastAccessTimestamp, which leads to an incorrect ledger handle invalidation.

Potential issue

The following issue still exists. I will push a new PR to fix it in the future.

time/thread	release idle ledger handle	ledgerHandle.readAsync
1	Ledger handle has been idle for a long time
2	Start to release ledger handle
3		a new reading request comes in
4	closed ledger handle
5		get a ledger closed error

Documentation

• doc
• doc-required
• doc-not-needed
• doc-complete

Matching PR in forked repository

PR in forked repository: x

[lhotari]

LGTM

[codecov-commenter]

Codecov Report

❌ Patch coverage is 66.66667% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 74.25%. Comparing base (bbc6224) to head (0a4eca2).
⚠️ Report is 1765 commits behind head on master.

Files with missing lines	Patch %	Lines
...che/bookkeeper/mledger/impl/ManagedLedgerImpl.java	50.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##             master   #24384      +/-   ##
============================================
+ Coverage     73.57%   74.25%   +0.67%     
+ Complexity    32624    32301     -323     
============================================
  Files          1877     1867      -10     
  Lines        139502   145268    +5766     
  Branches      15299    16610    +1311     
============================================
+ Hits         102638   107867    +5229     
+ Misses        28908    28863      -45     
- Partials       7956     8538     +582     

Flag	Coverage Δ
inttests	26.79% <0.00%> (+2.21%)	⬆️
systests	23.34% <50.00%> (-0.99%)	⬇️
unittests	73.73% <66.66%> (+0.89%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines	Coverage Δ
...oad/jcloud/impl/BlobStoreBackedReadHandleImpl.java	77.77% <100.00%> (-2.37%)	⬇️
...che/bookkeeper/mledger/impl/ManagedLedgerImpl.java	81.28% <50.00%> (+0.61%)	⬆️

... and 1080 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[lhotari]

• original: Pulsar only releases the offloaded ledger handle that is typed BlobStoreBackedReadHandle
• after [improve][ml]Release idle offloaded read handle only the ref count is 0 #24381, Pulsar will release the offloaded ledger handle immediately if its type is not BlobStoreBackedReadHandle, because the method lastAccessTimestamp() always returns -1 if its type is not BlobStoreBackedReadHandle

the title "[fix][ml]Revert a behavior change of releasing idle offloaded ledger handle: only release idle BlobStoreBackedReadHandle " is slightly misleading since even before it would only release idle BlobStoreBackedReadHandleImpls since that's the only implementation of OffloadedLedgerHandle.
Revisiting the PR title to focus on an improvement to avoid a remaining race condition could be better.