Various fixes

[bcipolli]

Core changes:

• Randomize the elongation of "patches" (was fixed to 1.5)
• Use the timestamp of the dataset in the model invalidation.
• Fix labels of RH/LH in 102x75 scripts.

Should not use the randomized elongation of patches until we verify results replicate... it is by default here.

[bcipolli]

These are simple changes; it just allows cells to vary in their elongation. Let's merge this and see if affects any of our results :)