chore: replace ObjectStore type requirement with ObspecInput

[hrodmn]

resolves #28

[github-actions]

Benchmark Comparison

Test	Baseline (ms)	PR (ms)	Change
test_band_access_pattern[multi_band]	1325.1	1339.2	+1.1%
test_band_access_pattern[single_band]	1202.6	1239.0	+3.0%
test_full_compute	627.2	658.4	+5.0%
test_mosaic_method[first]	613.3	617.8	+0.7%
test_mosaic_method[median]	641.4	680.7	+6.1%
test_native_crs_resolution	10757.3	11923.2	+10.8% ⚠️
test_open_overhead	270.5	298.9	+10.5% ⚠️
test_reproject_workers[1]	1354.0	1419.4	+4.8%
test_reproject_workers[4]	1348.5	1384.9	+2.7%
test_time_step_parallelism[dask_time_1]	1331.2	1332.8	+0.1%
test_time_step_parallelism[no_dask]	914.6	954.0	+4.3%

[kylebarron]

I think a link here would be good. You can link to https://developmentseed.org/async-geotiff/latest/api/geotiff/#async_geotiff.GeoTIFF.open, and then potentially also to the ObspecInput type

[kylebarron]

Hmm, perhaps we should re-export ObspecInput through the async-geotiff API too? 🤷‍♂️

I don't really like the name ObspecInput either. I think just Store would be more descriptive. Any thoughts?

[hrodmn]

It seems a little odd to be importing from async-tiff but it was already an indirect dependency so it's not a big deal.

Store would be a more intuitive name. ObspecInput says a lot about what the class is but not very much about what it does.

[kylebarron]

developmentseed/async-geotiff#161

With 0.5.1 you should be able to use the Store protocol defined in async-geotiff