Allow rerun selected test cases

[Raymond112514]

This PR adds functionality to the BFCLv3 dataset, allowing users to select specific test cases to run. Users can specify the selected test cases in a .txt file within the Berkeley Function Calling Leaderboard directory and run these cases by adding the --rerun argument:

bfcl generate --model MODEL_NAME --test-category TEST_CATEGORY --num-threads 1 --rerun FILE.txt

(Similarly for locally hosted models). For instance, FILE.txt can contain:

simple_0
simple_23
simple_45

Main change: added the overwrite function in base_handler, which takes in a newly executed result and replaces the old one in place.

[Raymond112514]

Added the --rerun-all flag. When this flag is present, the results are overwritten.
Changed the logic of collect_test_case slightly.

[Raymond112514]

Added the --result-dir and --score dir option.