Duplicate creation of index in join

[harshitgupta412]

Describe the bug
Every join to sem_join_cascade creates the index for second series (l2) from scratch since it call sem_index on it. Also, it can potentially override an existing index.

See:

lotus/lotus/sem_ops/sem_join.py

Line 287 in 9761855

l2_df = l2_df.sem_index(col2_label, f"{col2_label}_index")

Expected behavior
The index should be recreated only if the data has changed. Using time to differentiate the versions or creating it in ~/.cache might be better so that we don't override the user's existing indices.