Show actual image on hover in individual visit plot.

[dharhas]

This is an enhancement of the 'drill down' individual visit plots created in #110

On hovering on the skyplot? individual pixels we should show individual photos/pngs?

@timothydmorton can you elaborate.

[timothydmorton]

Not on hover; on click, and on both scatter and sky plots. I am assuming that we improve our datashaded images such that they transition to points when zoomed in far enough (I believe this was in the plan, right @philippjfr?). If that is done, then I would want a small thumbnail image (final content TBD by me) to be loaded into a tooltip-y thing (or even in another small panel somewhere in the layout if that's easier than a tooltip-y thing; open to suggestions @brendancol) upon click of a point.

[dharhas]

@philippjfr is working on the transition to points work. Let me add an issue for that

[philippjfr]

I am assuming that we improve our datashaded images such that they transition to points when zoomed in far enough (I believe this was in the plan, right @philippjfr?)

Indeed.

If that is done, then I would want a small thumbnail image (final content TBD by me) to be loaded into a tooltip-y thing (or even in another small panel somewhere in the layout if that's easier than a tooltip-y thing

Should be straightforward as long as the images are accessible somehow, if they're only accessible from the server we could base64 encode them, otherwise we can just link to them.

[timothydmorton]

The image data (large images) is stored on disk in known locations, and there are LSST stack routines for extracting cutout subsets of those large images. So the function I would write would return the cutout. The catch here is that each point would need to contain all the ID information necessary to retrieve the image, and to pass that id info in a dataID dictionary to the get_cutout() function. For coadds, sufficient information should be filter, tract, patch, ra, dec, and for visits, sufficient info should be filter, visit, ra, dec.