Emit (offset, length) information during generation

[joshmoore]

Goal would be to eventually generate a fsspec-reference-maker JSON file to allow for the consumption of OME-TIFFs as Zarr. I assume the information is available during the first pass through tile writing. If we can even just have that information in the log output, I should be able to generate the necessary JSON. Alternatively, raw2ometiff could generate the JSON itself e.g.:

$ conda create -n tifffile tifffile
$ conda activate tifffile
$ conda install -c ome/label/pre bioformats2raw raw2ometiff
$ bioformats2raw a.fake out
$ raw2ometiff out a.ome.tiff
$ tiff2fsspec a.ome.tiff file:///
$ cat a.ome.tiff.json
{
 ".zgroup": "{\n \"zarr_format\": 2\n}",
 ".zattrs": "{\n \"multiscales\": [\n  {\n   \"datasets\": [\n    {\n     \"path\": \"0\"\n    },\n    {\n     \"path\": \"1\"\n    }\n   ],\n   \"metadata\": {},\n   \"name\": \"a\",\n   \"version\": \"0.1\"\n  }\n ]\n}",
 "0/.zarray": "{\n \"chunks\": [\n  512,\n  512\n ],\n \"compressor\": {\n  \"id\": \"imagecodecs_lzw\"\n },\n \"dtype\": \"|u1\",\n \"fill_value\": 0,\n \"filters\": null,\n \"order\": \"C\",\n \"shape\": [\n  512,\n  512\n ],\n \"zarr_format\": 2\n}",
 "1/.zarray": "{\n \"chunks\": [\n  256,\n  256\n ],\n \"compressor\": {\n  \"id\": \"imagecodecs_lzw\"\n },\n \"dtype\": \"|u1\",\n \"fill_value\": 0,\n \"filters\": null,\n \"order\": \"C\",\n \"shape\": [\n  256,\n  256\n ],\n \"zarr_format\": 2\n}",
 "0/0.0": ["file:///a.ome.tiff", 6997, 47107],
 "1/0.0": ["file:///a.ome.tiff", 16, 6981]

see:

• https://github.com/intake/fsspec-reference-maker#version-1
• https://github.com/cgohlke/tifffile/blob/581d7a5d4d7784154066b9f11a0167bc08570b7c/tifffile/tiff2fsspec.py

[melissalinkert]

A place to start: #61

[melissalinkert]

See also feedback: #61 (comment)

[melissalinkert]

Closing as duplicate of #95.