Fix summary reading being too slow in case of many summary keys to match against

[eivindjahren]

Resolves #7262

looping over a fnmatch is too slow, especially after converting the fnmatch from c to python. Compiling a regex once is much faster, see the included test.

I ran the following test and the result of the new function seems to be identical to the old function.

from hypothesis import given, settings
from fnmatch import fnmatch, translate
import re
import hypothesis.strategies as st


def _fetch_keys_to_matcher(fetch_keys):
    if not fetch_keys:
        return lambda _: False
    regex = re.compile("|".join(translate(key) for key in fetch_keys))
    return lambda s: regex.fullmatch(s) is not None


def _should_load_summary_key(data_key, user_set_keys):
    return any(fnmatch(data_key, key) for key in user_set_keys)


@settings(max_examples=10000)
@given(st.lists(st.text()), st.text())
def test_are_the_same(fetch_keys, key):
    assert _fetch_keys_to_matcher(fetch_keys)(key) == _should_load_summary_key(
        key, fetch_keys
    )

I took some benchmarks using random but realistic fetch_keys and keys.

The new algorithm can sometimes perform worse when the size of fetch_keys is close to or smaller than the number of keys, but the old algorithm is regularly worse as the number of keys grow.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 88.88889% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 84.45%. Comparing base (2f02032) to head (e0c2692).

Files	Patch %	Lines
src/ert/config/_read_summary.py	88.88%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #7410      +/-   ##
==========================================
- Coverage   85.19%   84.45%   -0.75%     
==========================================
  Files         384      384              
  Lines       22860    22864       +4     
  Branches      882      883       +1     
==========================================
- Hits        19476    19309     -167     
- Misses       3276     3442     +166     
- Partials      108      113       +5     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.