Decouple background source in ResultsHandler

[mlincett]

In order to be able to load the background from a different analysis, the pickle directory structure can no longer being semi-statically defined at the ResultsHandler level, but is defined, starting from a base path, by means of a utility class PickleCache. Likewise, the entire managing logic (read/merge/write/clean) of the pickle cache is now encoded there.

A background_source parameter can be passed to the ResultsHandler constructor, so background data can be additionally loaded (and merged).

The solution is still not the cleanest, but works. Still at RFC stage.

[Copilot]

Pull Request Overview

This PR refactors the results‐handling logic by introducing a dedicated PickleCache class to manage pickle I/O and adds an optional background_source parameter to allow separate loading and merging of background trials.

• Introduced PickleCache to encapsulate pickle directory (read/merge/clean) logic.
• Added background_source to ResultsHandler to load and merge background data separately.
• Refactored ResultsHandler to use PickleCache and removed the legacy merge_pickle_data method.

[Copilot]

[nitpick] Remove or implement the commented-out directory creation logic in PickleCache.__init__ to avoid dead code.

Suggested change

	# self.pickle_path.mkdir(parents=True, exist_ok=True)
	self.path.mkdir(parents=True, exist_ok=True)

[Copilot]

There's an extra 'f' before {n_pending} causing the log to print 'f'; remove the 'f' so it reads {n_pending}.

Suggested change

	f"Appending f{n_pending} background data to {len(output_dict[background_label]['TS'])} existing trials ({scale_label=})"
	f"Appending {n_pending} background data to {len(output_dict[background_label]['TS'])} existing trials ({scale_label=})"

[Copilot]

Replace this print statement with a logger call (e.g., logger.info) to maintain consistent logging practices.

Suggested change

	print("NOTE!!!! Loading BG")
	logger.info("NOTE!!!! Loading BG")

[Copilot]

Replace this print statement with a logger call (e.g., logger.warning) for consistency and better control over output.

Suggested change

	print("NOTE!!!! No BG pickle cache")
	logger.warning("NOTE!!!! No BG pickle cache")

[Copilot]

Avoid calling sys.exit during initialization; consider raising an exception so callers can handle errors more gracefully.

Suggested change

	sys.exit()
	if not scale_shortener(0.0) in self.results:
	logger.error(
	f"No key equal to '0' in results! Keys are {self.results.keys()}"
	)
	sys.exit()
	raise RuntimeError("No data was found by ResultsHandler object!")
	if not scale_shortener(0.0) in self.results:
	logger.error(
	f"No key equal to '0' in results! Keys are {self.results.keys()}"
	)
	raise KeyError(f"No key equal to '0' in results! Keys are {self.results.keys()}")

[Copilot]

Avoid calling sys.exit during initialization; consider raising an exception so callers can handle errors more gracefully.

Suggested change

	sys.exit()
	if not scale_shortener(0.0) in self.results:
	logger.error(
	f"No key equal to '0' in results! Keys are {self.results.keys()}"
	)
	sys.exit()
	raise RuntimeError("No data was found by ResultsHandler object!")
	if not scale_shortener(0.0) in self.results:
	logger.error(
	f"No key equal to '0' in results! Keys are {self.results.keys()}"
	)
	raise RuntimeError("No key equal to '0' in results!")