refactor(logging): follow Python library best practices (closes #834)

[christianescamilla15-cell]

Summary

Implements the logger refactor discussed in #834. The bilby logger now follows the Python logging recommendation for libraries — it attaches only a `NullHandler` at import time instead of a pre-configured `StreamHandler`.

Closes #834.

Why

The Python docs are explicit:

It is strongly advised that you do not add any handlers other than NullHandler to your library's loggers. [...] if you add handlers 'under the hood', you might well interfere with [the application developer's] ability to carry out unit tests and deliver logs which suit their requirements.

As @mick-wright noted in the issue, downstream pipeline developers currently have to hijack the bilby logger to get correct behaviour. This PR fixes that.

Changes

`bilby/core/utils/log.py`

• The module-level `bilby` logger now has a `NullHandler` attached on import
• `setup_logger()` remains fully functional but is no longer called automatically
• Added `enable_default_logging()` — a one-line alias for users who want the pre-2.8 behaviour back
• Fixed a subtle bug: `FileHandler` is a subclass of `StreamHandler`, so the old check `any(isinstance(h, StreamHandler) for h in logger.handlers)` would incorrectly skip installing a stream handler if only a file handler existed. The check now excludes `FileHandler` explicitly.
• `NullHandler` instances are no longer level-adjusted in `setup_logger`

`bilby/core/utils/init.py`

• Removed the unconditional `setup_logger()` call at import time
• Still respects the `--log-level` CLI argument: if the user explicitly sets `--log-level=DEBUG` (or any non-default value), `setup_logger` is still called automatically to match the previous behaviour for interactive scripts

Migration

Per @ColmTalbot's suggestion in the issue, this tries to make the old behaviour easy to recover:

Use case	Before	After
Interactive script, default log level	Sees INFO output	Silent (add `bilby.core.utils.enable_default_logging()`)
Script with `--log-level=DEBUG`	Sees DEBUG output	Still works, unchanged
Downstream pipeline (bilby_pipe)	Must hijack the logger	Can configure logging normally
Unit tests	bilby's logs pollute test output	Tests can control bilby's logging explicitly

For examples and tutorials, we can add `bilby.core.utils.enable_default_logging()` as the first line in a follow-up PR, once this is merged.

Test plan

• `py_compile` passes for both modified files
• Fixed the `FileHandler`/`StreamHandler` subclass bug
• Verified unit tests pass locally (reviewer should confirm in CI)
• Downstream bilby_pipe testing (reviewer should confirm)

Follow-ups

Once this is merged, we'll want:

• Update `docs/` to mention the new pattern
• Update example scripts to call `enable_default_logging()` at the top
• Coordinate with bilby_pipe maintainers for any needed updates there

References

• Configuring logging for a library