Performance issue with LoadScopeScheduling when a node has a large number of tests

I've noticed the use of index() in this code block causes a significant performance issue when a node has a large number of tests. This can be improved by precomputing a mapping from node ID to index using a dictionary and look it up in the loop.

eg.

index_map = {nodeid: i for i, nodeid in enumerate(worker_collection)}
nodeids_indexes = [
    index_map[nodeid]
    for nodeid, completed in work_unit.items()
    if not completed
]

To reproduce the issue, you can run pytest -n 2 --dist=loadscope with the following test code:

import pytest

@pytest.mark.parametrize("i", range(100000))
def test_something(i):
    pass

Start in src/xdist/scheduler/loadscope.py around lines 274-280, where worker collection indexes are built. Reproduce with pytest -n 2 --dist=loadscope and the 100000-case parametrized test from the issue. Done means the scheduling path avoids the reported lookup slowdown while preserving the expected node ID indexes.