[Tilesathome] Proposal: New T at H Server structure

[Alan Millar]

>> Level 13 will allow most t at h clients to run complex tiles much faster,
>> creating a fast turnaround time. Stitching four captionless
>> tiles 13 will create a level 12 tile.

>> Is this a too big change of strategy ?

I think there are many places throughout the project that are dependent on
z12 being a key layer.  That won't be a simple change.

> Starting at level Z13 will immediately help.
> If , in addition we provide a captionless layer on level
> 12 AND 13 , the area that needs rendering outside the tile
> could also be reduced.

I think it could work if we have the client do z13 tiles, stitch them
together locally, and then upload the same things they upload now:  full
tiles for 12 through 17, and a captionless for just 12.

The server, and other later processes, won't care if the captionless z12
came from a single render or 4 stitched together.  Same thing with the
full tiles, as long as the captions are correct.

The client could request a z12, like now, and if the complexity is over
some threshold, run 4 z13s against it.  Or 16 z14s if needed.  Keep the
results and stitch them locally.

The main problem I imagine is large labels which might span across z13
tiles.  Perhaps a hybrid semi-captionless approach would work.  Render the
z13 tiles with way labels but not POI labels, and render a z12 caption
layer with POI labels and composite them.  All local to the t at h client;
only upload the final results like now.

This is sounding complicated, but could be possible without impacting all
other current z12 dependencies.

- Alan