…a Managed Instances (aws#219)

* feat: Emit worker pool size as a DEBUG log event during init on Lambda Managed Instances

* fix: Stringify non-serializable JSON log values and close init log sink

Address PR review feedback:
- Add default=str to the JSON log encoder so dict messages (and extra
  attributes) containing non-serializable values are stringified instead
  of raising and dropping the log record.
- Close the log sink opened by init_logging deterministically after the
  parent's handler is removed, instead of relying on GC.

* fix: Remove redundant output redirect from parent process

RAPID wires the runtime main process's stdout/stderr to the log egress
at spawn, so the parent does not need to dial the telemetry FD provider
socket — that socket exists for the forked workers, which continue to
redirect in run_single. Addresses PR review feedback.

…s#222)

Debian 11 bullseye's bullseye-security apt Release file has expired on
the mirror, causing 'apt-get update' to fail (exit 100) in the echo
Dockerfile and breaking all bullseye integration jobs regardless of
Python version. Bullseye is oldstable/EOL-track; replace it with the
current stable release, trixie (Debian 13), across Python 3.10-3.14.

Adapted from awslabs/aws-lambda-cpp replicate-source.yml:
- Branch guard changed from master to main
- Archive path and S3 object key namespaced under
  aws-lambda-python-runtime-interface-client

Update Python versions and add missing requirements

* feat(context): add w3c() helper for W3C trace fields

* feat: use python 3.14

* feat: use main for harness tests

* fix(ci): drop unused Set up Python step so Post cleanup stops failing

The dockerized-test workflow never runs pip on the runner — Python is only
invoked inside the Dockerfile build using the base image's own Python. The
'cache: pip' directive on setup-python@v5 therefore ran its Post cleanup
against an empty /home/runner/.cache/pip and failed the job even though all
12 dockerized suite cases passed.

Also bumps actions/checkout to v5 to drop the Node 20 deprecation warning.

This reverts commit fa006b6 (aws#226).

The pre-fork hooks change introduced in 4.1.0 is breaking, so it is
removed: the `awslambdaric.lambda_concurrency_hooks` module, the
`@register_pre_fork` decorator and the `Runtime.PreForkError` error
type are gone, restoring the 4.0.4 multi-concurrent startup flow. The
on-demand path is unaffected.

Bumps the version to 4.2.1 and adds a release note.

fix: revert pre-fork hooks for multi-concurrent mode

multiprocessing.Process used the interpreter default start method,
which is fork on 3.13 and forkserver on 3.14+, so the worker process
model differed by runtime. Pin "fork" via get_context so workers are
direct children of the runtime process on every version and the parent
default is left untouched for customer code.

fix: pin multi-concurrent worker start method to fork

feat: bump to 4.2.2