Store original_metadata.txt for all images in a plate at the plate level

[jamoore]

Discussion from devteam today as a response to if an original_metadata.txt object gets attached to all images in plate:

chris 16:54 
Josh: In a Plate context we don't store any original metadata.
andrew patterson left the room.
16:55
andrew patterson [ajpatterson@jabber.lifesci.dundee.ac.uk/adium] entered the room.
16:56
andrew patterson left the room.
16:56
josh moore 17:01 
chris: easy enough for the upgrade, any reason not to, though?
chris 17:02 
The reason why we don't do it?
josh moore 17:02 
yeah.
chris 17:02 
Because it's a complete nightmare and make the imports 2 or 3 times slower.
03:04
On a 384 well, 2 field plate that set of operations is 10+ minutes.
04:04
768 original files, 768 file annotations, 768 links, 768 interactions with original file store... etc. it's horrible.
05:06
It was so bad on some of the higher field count plates that it was actually taking longer to save those objects in the database than it took to copy the planes up.
josh moore 17:05 
ah, i was assuming there would have been one original_metadata.txt at the plate level.
chris 17:05 
Which is useless since every image is different.
josh moore 17:06 
i was also assuming "[image 4] key1=val [image 5] key1=..." etc.
chris 17:06 
We certainly could do that.
07:06
But it doesn't right now obviously.