Propagate copy_from through SubImage

[197g]

See #2766 but for all SubImage<I> that is improved. Gains from this are even more absurd, as previous code would unroll or elide much based on the in-bounds verification—complicating the indexing in the tight loop significantly.

Details

copy_subimage_from      time:   [530.38 µs 548.81 µs 565.15 µs]
                        change: [-92.413% -92.100% -91.797%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high severe

copy_subimage_at        time:   [39.675 µs 40.163 µs 40.747 µs]
                        change: [-95.561% -95.440% -95.324%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) high mild
  6 (6.00%) high severe

copy_subimage_view      time:   [42.345 µs 42.665 µs 43.014 µs]
                        change: [-94.854% -94.675% -94.494%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 2 outliers among 100 measurements (2.00%)
  1 (1.00%) high mild
  1 (1.00%) high severe

copy_subimage_fill      time:   [132.67 µs 133.60 µs 134.73 µs]
                        change: [-83.000% -82.719% -82.424%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 4 outliers among 100 measurements (4.00%)
  2 (2.00%) high mild
  2 (2.00%) high severe

copy_subimage_strides   time:   [36.394 µs 36.609 µs 36.860 µs]
                        change: [-81.753% -81.251% -80.794%] (p = 0.00 < 0.05)
                        Performance has improved.
Found 9 outliers among 100 measurements (9.00%)
  8 (8.00%) high mild
  1 (1.00%) high severe