makes the BOSS backend cuboid aware, and requests chunks in those sizes
uses caching for data chunks
adds some refactorings:
avoid searching lists in BOSS and neuroglancer backend, using dicts instead
moves _chunks and _cutout to Backend baseclass (maybe undoing this, if we don't align to cuboids in BOSS)
some minor stuff
~~Using the cuboids delayes the arrival of the first chunks heavily, which gives a bad UX
I would rather not use cuboid chunks here, and instead cache the chunks in the size of the incoming requests.~~
This PR:
makes the BOSS backend cuboid aware, and requests chunks in those sizesmoves_chunks
and_cutout
toBackend
baseclass (maybe undoing this, if we don't align to cuboids in BOSS)~~Using the cuboids delayes the arrival of the first chunks heavily, which gives a bad UX I would rather not use cuboid chunks here, and instead cache the chunks in the size of the incoming requests.~~
fixes #61