Support lossless float JPEG-XL

[kiwixz]

Is your feature request related to a problem? Please describe. I'm using a lossless floating point JPEG-XL to define some kind of Hald LUT. The goal here is to maximize quality, size is not a problem. Darktable seems to force a compression of at least distance 0.1f for libjxl < 0.9.0.

Describe the solution you'd like I'd like Darktable to consider a quality of 100 as lossless even for floating-point JPEG-XL. This is also what cjxl -q 100 test.pfm is doing. I understand that floating-points are inherently approximate, but I just want to tell libjxl to not mess even more with my pixels.

Alternatives Relying on distance 0 being less harmful with libjxl >= 0.9.0. Not sure if that also overrides the internal/original profile.

Additional context https://github.com/darktable-org/darktable/blob/a93d38c4db020fc66bdd97cb1abc1b2a02366da1/src/imageio/format/jxl.c#L161-L181