Integrate and update ASV based benchmarking suite

[ivirshup]

Integrate the benchmark suite started in https://github.com/ivirshup/anndata-benchmarks into this package.

Initial Goals

• [ ] Benchmarks for all basic operations
• [ ] Documentation on how to run and use benchmarks
• [ ] Parameterization strategy for benchmarks

Eventual goals

• [x] Dedicated benchmarking machine
• [ ] CI job for benchmarking

[ilan-gold]

More granular dev goals:

• [ ] Individual benchmarks for individual datasets i.e., smaller ones for local dev, bigger ones for the CI jobs on the machine
• [ ] Parametrization of the benchmarks based on e.g., dataset size or properties
• [ ] Define other "granular operations" (cc: @ivirshup; sparse dataset getting is a must and covered, but we probably want things like i/o, concatenation, merging etc. as well tested)

[ilan-gold]

Relatedly @flying-sheep does the machine include some sort of storage so we could cache big datasets?