Additional optimizations for similar_variables.py

[0xGusMcCrae]

Issue #1630

Adding in the additional optimizations for the similar variable name detector as described in #1945 (comment)

Changes:

• length check brought out of similar into detect_sim. If the lengths aren't the same, no need to run v2.name.lower()
• v1.name.lower() and v2.name.lower() are cached in detect_sim and given as inputs for similar so that .lower() doesn't have to be run an extra 2 times in the SequenceMatcher call.

Again running on the C4 contest Maia codebase:

Current slither/dev:

ncalls  tottime  percall  cumtime  percall filename:lineno(function)

40      0.004    0.000    17.769   0.444   .../slither/detectors/variables/similar_variables.py:83(_detect)

With new optimizations:

ncalls  tottime  percall  cumtime  percall filename:lineno(function)

40      0.004    0.000    14.867   0.372   .../slither/detectors/variables/similar_variables.py:84(_detect) 

Detected issues unchanged, ~16% reduction in runtime.

[montyly]

This is great, thanks @0xGusMcCrae for all these optimizations