xl/trax: search_tracks: remove time.sleep(0) (#844)

When processing the tracks in `search_tracks` generator, do not
explicitly yield the current thread's time-slice via
`time.sleep(0)`.

Removal of sleep call prevents unnecessary slow-downs in the main
thread, where we want to finish the operation as fast as possible
(especially under python 3.11 on linux, where `time.sleep(0)`
introduced additional overhead).

The comment accompanying the sleep call refers to GIL behavior,
which might have been applicable in python2-era, but does not
seem to be applicable to python3 anymore.

xl/trax/search.py

@@ -24,7 +24,6 @@
 # do so. If you do not wish to do so, delete this exception statement
 # from your version.
 
-import time
 import re
 from typing import Collection
 
@@ -537,14 +536,6 @@ def search_tracks(trackiter, trackmatchers: Collection[TracksMatcher]):
         if all(tma.match(srtr) for tma in trackmatchers):
             yield srtr
 
-        # On large collections, searching can take a while. Due to
-        # peculiarities in python's GIL that means the now-cpu-bound
-        # thread running the search can end up blocking other threads.
-        # Calling out to time.sleep forces a release of the GIL and
-        # allows other threads to run. Benchmarks show this has no
-        # noticeable effect on search speed.
-        time.sleep(0)
-
 
 def search_tracks_from_string(
     trackiter, search_string, case_sensitive=True, keyword_tags=None