Benchmarking as part of pytest

[ngupta23]

Add pytest runs for benchmarking (may have to be a limited set given the time constraints) similar to what is done in statsforecast.

https://github.com/Nixtla/statsforecast/blob/d7bceb07e5eda3a74d638dffaaf2023d0b8f8c15/.github/workflows/ci.yaml#L119
https://github.com/Nixtla/statsforecast/blob/main/experiments/m3/src/evaluation.py

[ngupta23]

Question: Should this be done here or in the pycaret repo?