perf: comment_by_id

[faretek1]

see issue #553

Changes

edits the User.comment_by_id logic to use an exponential-binary search to get comments faster.

Caution

This does not work right now! This is because there is a great number of edge cases that must be considered.

1. consider empty pages (these are not necessarily the last)
2. consider error 404 pages (these are necessarily beyond the end
3. consider the fact that comments can move between pages (but only forward) - this is noticeable with griffpatch's page
4. make sure the logic works for comments at the very start and end of pages
5. you will want to transfer these changes to studios and projects...right?

and then the behaviors that correspond to them:

1. don't immediately exit, and check forwards/backwards linearly until a comparison can be made. when both sides are reached, evaluate whether the comemnt_id is in this range; if so, comment not found
2. check backwards linearly? it tends to be page 67 for some reason but idk why
3. This one is the hard one
4. use <, >, <=, and >= appropriately and test them well
5. reimplement twice or create a layer of abstraction that can be used (interface?)

Tests

test_comment_by_id.py. In production, we will need to test this on comments on scratchattachV2