Optimise ranges.any_overlapping

[Peter554]

This PR optimises the ranges.any_overlapping function. It uses pytest-benchmark to measure the improvement. The resulting function is >100x faster for this benchmark..!

Kraken: This function is often used to determine whether there are overlaps in meter readings etc and has a significant impact on costing performance.

-------------------------------------------------------------------------------------------- benchmark: 2 tests --------------------------------------------------------------------------------------------
Name (time in ms)                            Min                 Max                Mean            StdDev              Median               IQR            Outliers       OPS            Rounds  Iterations
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_any_overlapping (NOW)                1.4645 (1.0)        1.5542 (1.0)        1.4730 (1.0)      0.0099 (1.0)        1.4699 (1.0)      0.0032 (1.0)         70;93  678.8814 (1.0)         641           1
test_any_overlapping (0004_ce35b5e)     191.0346 (130.44)   207.6654 (133.61)   194.7505 (132.21)   6.6148 (666.40)   191.3241 (130.16)   4.8863 (>1000.0)       1;1    5.1348 (0.01)          6           1
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

[romainletendart]

Looks good to me, well done! 👏

[romainletendart]

Suggested change

	- Optimise the `ranges.any_overlapping` function [#184](https://github.com/octoenergy/xocto/pull/184).
	- Improve the performance of the `ranges.any_overlapping` function (with some benchmark showing a >100x speed up) [#184](https://github.com/octoenergy/xocto/pull/184).

or any similar phrasing would give a better idea of the actual change.

[ngnpope]

The tests don't cover non-consecutive overlapping ranges which is important to highlight that this sorted() call is necessary.

class TestAnyOverlapping:
    ...

    @pytest.mark.parametrize(
        "ranges_",
        [
            [
                ranges.Range(0, 2),
                ranges.Range(4, 5),  # Added this to ensure the overlapping ranges are not adjacent.
                ranges.Range(1, 3),
            ],
            [
                ranges.Range(
                    0, 2, boundaries=ranges.RangeBoundaries.INCLUSIVE_INCLUSIVE
                ),
                ranges.Range(2, 4),
            ],
        ],
    )
    def test_returns_true_if_and_ranges_overlap_non_consecutive(self, ranges_):
        assert ranges.any_overlapping(ranges_)

[Peter554]

Yes nice, added

[ngnpope]

We don't need to keep a stack as we're only looking at the last item:

Suggested change

	stack: list[Range[T]] = []
	for range in sorted(ranges):
	if len(stack) == 0:
	stack.append(range)
	else:
	if stack[-1].intersection(range):
	return True
	else:
	stack.append(range)
	prev_range: Range[T] | None = None
	for range in sorted(ranges):
	if prev_range is not None and prev_range & range:
	return True
	prev_range = range

[Peter554]

True! That even helped the performance a little more 👍