Complete dtype feature support

[fm3]

This table should track the current state of wk dtype support.

dtype	color layers	segmentation layers	histogram	thumbnail
uint8	✓	✓	✓	✓
uint16	✓	✓	✓	✓
uint24 rgb	✓	does not apply	✓	✓
uint32	✗	✓	✓	✓
uint64	✗	(✓) (til 2⁵³−1)	✓	✓ (segmentation only)
int8	✗	✗	✗	✗
int16	✗	✗	✗	✗
int32	✗	✗	✗	✗
int64	✗	✗	✗	✗
float	✓	✗	✓	✗
double	✗	✗	✗	✗

[normanrz]

Can we support this?

dtype	color layers	segmentation layers	histogram	thumbnail
uint8	✓	✓	✓	✓
uint16	✓	✓	✓	✓
uint24 rgb	✓	does not apply	✓	✓
uint32	☑️	✓	✓	✓
uint64	✗	✓ (til 2⁵³−1)	✓	✓ (segmentation only)
int8	☑️	☑️	☑️	☑️
int16	☑️	☑️	☑️	☑️
int32	☑️	☑️	☑️	☑️
int64	✗	☑️	✗	✗
float	✓	✗	✓	☑️
double	✗	✗	✗	✗

[fm3]

Signed integers for segmentations may lead to some pitfalls. We’d have to either prevent or deal with negative Segment IDs. I think there are a lot of things relying on all segment IDs being positive.