MicroManager: use PositionName key/value for storing the ImageName

[sbesson]

See https://trello.com/c/5KT3Bvie/1-idr0040-aymoz-singlecell

Currently the Micro-Manager reader uses one of the TIFF filename for populating the ImageName metadata. For typical replica experiments across multiple positions (Pos0, Pos1,...), this results in duplicate image names. This commit now parses the PositionName key/value in the metadata JSON and use it if available to fill the ImageName.

This PR will probably affect the metadata of existing filesets, require a configuration change in which case it might be more suitable for a minor release increment.

/cc @dominikl

[sbesson]

With the last commit, https://ci.openmicroscopy.org/job/BIOFORMATS-DEV-merge-repository-subset/552/ indicates the following changes

sbesson@ls30630:tmp $ grep FAILED consoleText 
   [testng] [2018-03-01 08:12:55,441] [pool-1-thread-3] 	ImageNames: FAILED (Series 0 (got 'Pos1', expected 'img_000000000_00-BF0_000.tif'))
   [testng] [2018-03-01 08:12:55,520] [pool-1-thread-4] 	ImageNames: FAILED (Series 0 (got 'Pos0', expected 'img_000000000_00-BF0_000.tif'))
   [testng] [2018-03-01 08:13:44,663] [pool-1-thread-2] 	ImageNames: FAILED (Series 0 (got 'Pos0', expected 'Test_Dataset_MultiPosition_Stack_1_MMStack_Pos0_metadata.txt'))
   [testng] [2018-03-01 08:14:05,609] [pool-1-thread-1] 	ImageNames: FAILED (Series 0 (got '1-Pos_000_001', expected 'hyper2.0_1_MMStack_1-Pos_000_001_metadata.txt'))
   [testng] [2018-03-01 08:14:32,532] [pool-1-thread-4] 	ImageNames: FAILED (Series 0 (got 'Pos3', expected 'img_000000000_00-BF0_000.tif'))
   [testng] [2018-03-01 08:14:51,017] [pool-1-thread-3] 	ImageNames: FAILED (Series 0 (got '1-Pos_001_001', expected 'hyper2.0_1_MMStack_1-Pos_001_001_metadata.txt'))
   [testng] [2018-03-01 08:14:51,888] [pool-1-thread-3] 	ImageNames: FAILED (Series 0 (got '1-Pos_001_000', expected 'hyper2.0_1_MMStack_1-Pos_001_000_metadata.txt'))
   [testng] [2018-03-01 08:15:07,042] [pool-1-thread-4] 	ImageNames: FAILED (Series 0 (got 'Pos5', expected 'img_000000000_00-BF0_000.tif'))
   [testng] [2018-03-01 08:15:20,726] [pool-1-thread-4] 	ImageNames: FAILED (Series 0 (got 'Pos4', expected 'img_000000000_00-BF0_000.tif'))
   [testng] [2018-03-01 08:15:33,044] [pool-1-thread-4] 	ImageNames: FAILED (Series 0 (got 'Pos2', expected 'img_000000000_00-BF0_000.tif'))
   [testng] [2018-03-01 08:15:45,439] [pool-1-thread-4] 	ImageNames: FAILED (Series 0 (got 'Pos0', expected 'MMStack_Pos0_metadata.txt'))
   [testng] [2018-03-01 08:15:53,740] [pool-1-thread-1] 	ImageNames: FAILED (Series 0 (got '1-Pos_000_000', expected 'hyper2.0_1_MMStack_1-Pos_000_000_metadata.txt'))
   [testng] [2018-03-01 08:15:54,668] [pool-1-thread-1] 	ImageNames: FAILED (Series 0 (got 'Pos0', expected 'img_000000000__000.tif'))
BUILD FAILED

Excluding for now to reduce the impact on other Bio-Formats work and will discuss with @dominikl later today. Leaving @dgault and @melissalinkert to comment on whether they would think such a change would be acceptable on the mainline Bio-Formats to minimize potential further divergences with the IDR branch.

[sbesson]

--rebased-to #3092

[melissalinkert]

This seems fine to me, and the configuration changes make sense. Only question would be if adding the name field to Position affects memo files, but I guess this will become obvious via #3092?

[sbesson]

Thanks @melissalinkert, local testing suggested the memo files were invalidated by the new field addition. bbd2fb8 should reduce the impact by making the name transient.

[dgault]

With bbd2fb8 the existing memo files are unaffected and this is suitable for a patch release.

Tested manually using sample datasets which are affected based on https://github.com/openmicroscopy/data_repo_config/pull/274/files. The files continue to be read and displayed correctly, the only metadata change seen is to the image name which now matches the position name when available. This makes more sense than having duplicate image names using the filename.

Builds and tests are green with both this and the associated configuration PR included.