Strengthen retry repr, wait cap and idle time tests

[xuu33030]

Summary

Closes #715. This strengthens the tests; it does not fix or change production behavior.

• Assert controller class, configuration and name in repr, including asyncio and Tornado's sleep functions, without pinning memory addresses.
• Cover wait_incrementing(max=...) at and beyond the cap, for numeric and timedelta limits.
• Replace all four idle_for: mock.ANY expectations with exact totals. The existing wrappers actually use the default zero wait, so the successful retry cases now temporarily use wait_fixed(0.25) with only sleeping mocked out. The real retry loop accumulates 0.5s (sync) / 0.25s (async); stopping after the first attempt remains 0.
• Remove the two legacy string-strategy constructor smoke tests: they neither execute retries nor establish support for callable-incompatible strings. Existing callable-strategy tests remain.
• Add a release note as requested by the contribution guide.

Verification

On macOS, Python 3.10.20 (the minimum supported version), using the project's development dependencies and poe all in a network-denied sandbox:

• Ruff lint and format checks passed.
• Mypy passed: 19 source files.
• Pytest: 183 passed.
• Sphinx doctests: 35 passed.
• Sphinx HTML build passed with -W.
• git diff --check passed.

I also ran the full suite in separate temporary source copies, applying each mutation independently to both baseline (3e58094) and updated tests:

Deliberate production mutation	Baseline tests	Updated tests
BaseRetrying.__repr__ returns ""	Pass (missed)	Three repr tests fail
wait_incrementing omits min(result, self.max)	Pass (missed)	Both max-limit subtests fail
Statistics skip accumulating sleep into idle_for	Pass (missed)	Sync and async statistics tests fail

None of these mutations are included in the commit. Tests without mutations pass. Other Python versions/platforms have not been run locally and remain for CI.

AI assistance: this contribution was investigated, authored and locally verified using Hermes Agent.