Add multiple LINDI benchmarks to method rankings

[stephprince]

We had previously added multiple LINDI benchmarks to both make fair comparisons with some of the other streaming methods and measure performance in what would be a common use case.

• LindiLocalJSONFileReadBenchmark.time_read_lindi_pynwb: measures the time to open a LINDI file with pynwb. Demos what would be the common use case but is slightly different than other read benchmarks in that the LINDI file has been downloaded locally.
• LindiDandiAPIDownloadBenchmark.time_download_lindi_dandi_api: measures the time to download a LINDI file. Adding this time to the read benchmark above may provide a more fair comparison with some of the other benchmarks tests.

Currently the methods ranking heat map only includes the time_read_lindi_pynwb benchmark, but @oruebel and I had discussed adding both read and download + read to the final figure for the paper. The download + read measurements could be calculated from the current results by combining the two benchmarks above, but I think an even simpler approach to avoid additional LINDI-specific data transformations would be to add another benchmark test that times the combined operations.

[stephprince]

Note that adding these new benchmarks and including them in the method rankings may require adding to the DEFAULT_BENCHMARK_ORDER variable which dictates which methods will be included in the final figures:

nwb_benchmarks/src/nwb_benchmarks/database/_visualization.py

Lines 16 to 26 in d583edc

	DEFAULT_BENCHMARK_ORDER = [
	"hdf5 h5py remfile no cache",
	"hdf5 h5py fsspec https no cache",
	"hdf5 h5py fsspec s3 no cache",
	"hdf5 h5py remfile with cache",
	"hdf5 h5py fsspec https with cache",
	"hdf5 h5py fsspec s3 with cache",
	"hdf5 h5py ros3",
	"lindi h5py",
	"zarr s3",
	"zarr s3 force no consolidated",