feat(storage): Handle large read ranges in bidi read.

[bajajneha27]

Part of #15340

This change is 

[codecov]

Codecov Report

❌ Patch coverage is 95.65217% with 11 lines in your changes missing coverage. Please review.
✅ Project coverage is 92.99%. Comparing base (59e0f6b) to head (de3a436).
⚠️ Report is 2 commits behind head on main.

Files with missing lines	Patch %	Lines
...d/storage/internal/async/object_descriptor_impl.cc	89.47%	4 Missing ⚠️
...oud/storage/tests/async_client_integration_test.cc	93.22%	4 Missing ⚠️
...rnal/async/object_descriptor_connection_tracing.cc	0.00%	2 Missing ⚠️
google/cloud/storage/testing/mock_resume_policy.h	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #15152      +/-   ##
==========================================
+ Coverage   92.97%   92.99%   +0.02%     
==========================================
  Files        2402     2403       +1     
  Lines      218287   218506     +219     
==========================================
+ Hits       202961   203210     +249     
+ Misses      15326    15296      -30     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[devbww]

Reviewable status: 0 of 15 files reviewed, 6 unresolved discussions (waiting on @bajajneha27, @ddelgrosso1, and @scotthart)

---

google/cloud/storage/internal/async/object_descriptor_impl.cc line 251 at r4 (raw file):

Previously, scotthart (Scott Hart) wrote…

If you can guarantee that no other thread would ever be executing any of the code sections that modify streams_ at the same time that the current thread reads streams_ here, then you are ok without the lock.

And the only way to guarantee that is to show that another lock serializes these actions (in which case this lock isn't needed), or to show the actions are all run by the same thread (in which case this lock also isn't needed). Both of these require meta information, so the reliable way to make the guarantee locally is with a lock that protects both reads and writes.

[bajajneha27]

Reviewable status: 0 of 15 files reviewed, 6 unresolved discussions (waiting on @ddelgrosso1, @devbww, @rajeevpodar, and @scotthart)

---

google/cloud/storage/internal/async/object_descriptor_impl.cc line 251 at r4 (raw file):

Previously, devbww (Bradley White) wrote…

And the only way to guarantee that is to show that another lock serializes these actions (in which case this lock isn't needed), or to show the actions are all run by the same thread (in which case this lock also isn't needed). Both of these require meta information, so the reliable way to make the guarantee locally is with a lock that protects both reads and writes.

Done.

[bajajneha27]

I'm not sure why other jobs are not kicking in. There's nothing in the logs.

[scotthart]

Reviewed 3 of 7 files at r1, 4 of 7 files at r2, 7 of 8 files at r4, 1 of 1 files at r5, all commit messages.
Reviewable status: all files reviewed, 6 unresolved discussions (waiting on @bajajneha27, @ddelgrosso1, @devbww, and @rajeevpodar)

[scotthart]

/gcbrun