Expose STAC temporal metadata

[pomadchin]

Overview

This PR adds exposes stac temporal extent thought the temporal metadata (and covers #259 at the same time):

It also fixes #260:

• (works) http://localhost:9000/?SERVICE=WMS&VERSION=1.3.0&REQUEST=GetMap&BBOX=43.81618476019971808%2C-79.40248489379884234%2C43.94325769600655462%2C-79.15603855282134305&CRS=EPSG%3A4326&WIDTH=1103&HEIGHT=570&LAYERS=markham&STYLES=&FORMAT=image/png&DPI=72&MAP_RESOLUTION=72&FORMAT_OPTIONS=dpi%3A72&TRANSPARENT=TRUE&time=1960-01-01T12:00:00Z
• (returns an empty tile) http://localhost:9000/?SERVICE=WMS&VERSION=1.3.0&REQUEST=GetMap&BBOX=43.81618476019971808%2C-79.40248489379884234%2C43.94325769600655462%2C-79.15603855282134305&CRS=EPSG%3A4326&WIDTH=1103&HEIGHT=570&LAYERS=markham&STYLES=&FORMAT=image/png&DPI=72&MAP_RESOLUTION=72&FORMAT_OPTIONS=dpi%3A72&TRANSPARENT=TRUE&time=1961-01-01T12:00:00Z

Checklist

• Description of PR is in an appropriate section of the CHANGELOG and grouped with similar changes if possible

Closes #259
Closes #260
Closes https://github.com/azavea/geotrellis/issues/278

[CloudNiner]

A couple clarifying questions before a more thorough review to help me understand the changes made here

[CloudNiner]

Unless there's a really compelling reason otherwise, we should exhaustive match our ADTs. Also should fix up the regression in parsing params noted previously.

I'm willing to let the ADT change go through, but I think it's a bad idea, see comment for a few potential alternative solutions.

[pomadchin]

@CloudNiner the current OgcTime tree is the following:

sealed trait OgcTime
case object OgcTimeEmpty extends OgcTime
final case class OgcTimePositions(list: NonEmptyList[ZonedDateTime]) extends OgcTime
final case class OgcTimeInterval(start: ZonedDateTime, end: Option[ZonedDateTime], interval: Option[String]) extends OgcTime

And you think using OgcTimeInterval for everything would be a better idea?

1. Talking into account that both WMS and WCS services can work with both
2. Query evaluation in case of the OgcTimePositions works as the exact match, while OgcTimeInterval works as a search function on the time range, see link

[pomadchin]

I'll work on the interfaces a bit more, but I think that OgcTimeInterval should represent an interval where the OgcTimePositions should represent a list of times that are can not be reprepsented as an interval. Using the List[OgcTimeInterval] sounds like a semantic issue. How would I know how to print such list in the WMS capabilities?

[pomadchin]

As an alternative we can have a newtype List[OgcTimeInterval] which would have a configurable printer.

[CloudNiner]

We chatted a bit about this after standup, adding some notes here.

I like the idea of making OgcTimeInterval explicitly an interval and seeing how that falls out, which would make end required in the case class, not sure if interval can stay an option or not. Maybe we can avoid the List[OgcTimeInterval] case for now and continue to just handle a single OgcTimeInterval? IIRC printing an OgcTimeInterval in the response depends on the spec for each of the response formats.

[CloudNiner]

LGTM once the couple remaining minor issues are addressed and the build is green.

[CloudNiner]

😭