Fix tplstart values

[JenniferKarr]

made the tplstart values more reasonable - they now match the observtion date.

fixed a wrong template value in recipes.yaml

regenerated the data based on these values and regenerated checksums.

[hugobuddel]

All exposures from the same template should have the same TPL.START; is that the case with this PR?

In particular, the ESO stack (EDPS) by default groups data on TPL.START. So for example to (automatically) stack 10 raw darks into a master dark, then all raw darks should have the same TPL.START. (It is also possible to define different grouping rules, but we can stick to the default most of the time.)

[JenniferKarr]

At the moment, almost all the files have a single image per template (the minimum set needed for workflow generation); the exception is pairs of sci/std and sky (which for the time being have the same tpl.start). We don't currently consider template structure in the METIS_Simulations code, so implementing that will take a bit of thought and will not be done for Friday. The current state is more plausible than it was a couple of hours ago, however, when all the tpl.start values were identical.

[hugobuddel]

OK good, if we have only one file per template anyway, then this is fine. I see you fixed this in #73 so I'll close this, thanks!