Tracking issue for benchmarking improvements.

[prady0t]

• Delete asv benchmarks if any.
• Make pytest-codspeed benchmarks deterministic.
• Use pytest fixtures wherever possible instead of generating them inline.
• The benchmark suite needs to be reduced, perhaps by using hypothesis to sample from the overall set of benchmarks; and skipping the small datasets.
• Expand by benchmarking more functions (function's list to be added later).