Benchmark with pytest-benchmark

[dcherian]

• figure out some good groups to apply
• Post-processing script to do a useful compare :/
• make setup less painful. We should be able to call setup_benchmarks --extra-prefix=REF once and just reuse that to re-generate values.
• make sure we have a benchmark that captures improvements between alpha.8 and alpha.10

---

Experimenting with pytest-benchmark Here is some quite verbose output

------------------------------------------------------------------------------------------------------------------ benchmark: 5 tests ------------------------------------------------------------------------------------------------------------------
Name (time in us)                                        Min                       Max                      Mean                  StdDev                    Median                     IQR            Outliers         OPS            Rounds  Iterations
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_time_zarr_open[era5-single]                    679.6250 (1.0)          1,493.7920 (1.0)            812.5706 (1.0)          215.8252 (1.0)            744.2500 (1.0)           74.2080 (1.0)           1;2  1,230.6623 (1.0)          13           1
test_time_zarr_members[era5-single]               1,419.1250 (2.09)         4,540.1250 (3.04)         1,721.0148 (2.12)         605.7372 (2.81)         1,502.5840 (2.02)          95.6037 (1.29)        12;27    581.0525 (0.47)        135           1
test_time_xarray_open[era5-single]              163,319.3750 (240.31)     171,454.9160 (114.78)     167,938.4166 (206.68)     2,208.6063 (10.23)      167,957.1455 (225.67)     2,890.7910 (38.96)         2;0      5.9546 (0.00)         10           1
test_time_create_store[era5-single]             253,562.7920 (373.09)     273,780.2080 (183.28)     263,209.5416 (323.92)     7,906.1436 (36.63)      264,329.3750 (355.16)    11,918.9787 (160.62)        2;0      3.7993 (0.00)          5           1
test_time_xarray_read_chunks[era5-single]     5,404,355.4170 (>1000.0)  6,491,509.0410 (>1000.0)  5,876,143.3625 (>1000.0)  301,326.2796 (>1000.0)  5,869,075.6460 (>1000.0)  362,416.4170 (>1000.0)       3;0      0.1702 (0.00)         10           1
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Legend:
  Outliers: 1 Standard Deviation from Mean; 1.5 IQR (InterQuartile Range) from 1st Quartile and 3rd Quartile.
  OPS: Operations Per Second, computed as 1 / Mean