[DRAFT] feat: added client side metric instrumentation to read_rows and mutate_rows

[daniel-sanche]

Migration of googleapis/python-bigtable#1256 to the monorepo

Builds off of #16712 to add instrumentation to read_rows and mutate_rows, along with the mutation batcher

[gemini-code-assist]

Code Review

This pull request integrates client-side metrics tracking into Bigtable data operations, including ReadRows and MutateRows, for both asynchronous and synchronous implementations. It introduces tracked_retry to capture operation metrics and refactors read_row to utilize operation classes directly, supported by extensive new system and unit tests. Feedback identifies opportunities to improve the efficiency of read_row by using next() or __anext__ on generators instead of converting them to lists, and suggests adding a type ignore comment in the synchronous code for consistency.

[gemini-code-assist]

This implementation consumes the entire generator into a list just to get the first element. A more efficient and idiomatic way to get the first item from an async generator is to use anext.

Suggested change

	try:
	results = [a async for a in results_generator]
	return results[0]
	except IndexError:
	return None
	try:
	return await results_generator.__anext__()
	except StopAsyncIteration:
	return None

[gemini-code-assist]

The corresponding async version of this code includes a # type: ignore comment here. To maintain consistency and prevent potential mypy errors, it would be good to add it here as well.

Suggested change

	self._operation_metric.active_attempt.application_blocking_time_ns += (
	time.monotonic_ns() - block_time
	)
	self._operation_metric.active_attempt.application_blocking_time_ns += (
	time.monotonic_ns() - block_time
	) # type: ignore

References

1. Using type: ignore is an acceptable trade-off to satisfy mypy when precise type hints are not cost-effective or to maintain consistency with other parts of the codebase.

[gemini-code-assist]

This implementation consumes the entire generator into a list just to get the first element. A more efficient and idiomatic way to get the first item from a generator is to use next().

Suggested change

	try:
	results = [a for a in results_generator]
	return results[0]
	except IndexError:
	return None
	try:
	return next(results_generator)
	except StopIteration:
	return None