fix[next]: fix metrics collection when compiling the same program multiple times

[egparedes]

Refactor the instrumentation metrics to support multiple compiled program pools derived from the same program definition and backend.

Changes:

• Introduced pool-specific metrics tracking with a counter mechanism to distinguish between multiple pools with the same (program_name, backend_name) combination
• Refactored hook signatures to pass program_pool objects instead of individual compiled programs, enabling better metrics association
• Renamed metrics API functions for clarity: metrics_context → metrics_context_key, metrics_setter_at_enter → metrics_context_key_at_enter

[Copilot]

Pull request overview

This pull request refactors the metrics instrumentation system to support multiple compiled program pools derived from the same program definition and backend. The key changes include:

Changes:

• Renamed metrics API functions for clarity: metrics_context → metrics_context_key, metrics_setter_at_enter → metrics_context_key_at_enter
• Introduced pool-specific metrics tracking with a counter mechanism to distinguish between multiple pools with the same (program_name, backend_name) combination
• Refactored hook signatures to pass program_pool objects instead of individual compiled programs, enabling better metrics association

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
src/gt4py/next/instrumentation/metrics.py	Renamed context manager functions for clarity
src/gt4py/next/otf/compiled_program.py	Added pool counter and metrics source key generation, updated hook signatures
src/gt4py/next/ffront/decorator.py	Updated to use renamed metrics API
tests/next_tests/unit_tests/instrumentation_tests/test_metrics.py	Updated test to use renamed API
tests/next_tests/integration_tests/feature_tests/instrumentation_tests/test_hooks.py	Updated hooks and tests to match new signatures

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 6 out of 6 changed files in this pull request and generated 7 comments.

Comments suppressed due to low confidence (1)

src/gt4py/next/otf/compiled_program.py:130

• The docstring for compiled_program_call_context still documents parameters (compiled_program, root) that are no longer in the function signature, and doesn’t describe program_pool. Please update the Args section to match the new signature to avoid confusing hook implementers.

    """
    Hook called at the beginning and end of a compiled program call.

    Args:
        compiled_program: The compiled program being called.
        args: The arguments with which the program is called.
        kwargs: The keyword arguments with which the program is called.
        offset_provider: The offset provider passed to the program.
        root: The root of the compiled programs pool this program belongs to, i.e. a tuple of
            (program name, backend name).
        key: The key of the compiled program in the compiled programs pool.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 7 out of 8 changed files in this pull request and generated 4 comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[tehrengruber]

Suggested change

	# Only used to provide a more meaningful name to the metrics source key
	_pools_per_root: collections.Counter = collections.Counter()
	#: Counter that uniquely identifies a program pool of a program
	# When multiple instances of the same program with the same root exist
	# (e.g. same backend, but different backend options) each instance has its own
	# pool. This counter identifies them and e.g. allows us to give a more meaningful name
	# to the metrics source key.
	_pools_per_root: collections.Counter = collections.Counter()