refactor attribute and remove TMS from BaseReader init, and remove min/maxzoom from info response

[vincentsarago]

closes #744

This PR does:

• refactors the SpatialMixin base class

  • removes tms attribute
  • removes geographic_crs attribute
  • removes minzoom and maxzoom properties
  • add get_zooms(tms: TileMatrixSet) method
  • change geographic_bounds property to method geographic_bounds(crs: CRS)
  • update tile_exists and add tms argument

• changes .tile() method and add tms argument

• refactors the info() method

  • return dataset bounds (instead of geographic bounds)
  • add crs metadata (epsg or wkt)

• removes models.SpatialInfo and models.Bounds

• removes minzoom/maxzoom attributes in MultiBaseReader and MultiBandReader

• starts a migration guide

# before
import morecantile
from rio_tiler.io import Reader

tms = morecantile.tms.get("WebMercatorQuad")
with Reader("cog.tif", tms=tms, geographic_crs=tms.rasterio_geographic_crs) as src:
    _ = src.tile_exists(0, 0, 0)
    _ = src.tile(0, 0, 0)
    _ = src.minzoom
    _ = src.maxzoom
    _ = src.geographic_bounds

    info = src.info()
    assert info.bounds == src.geographic_bounds
    assert info.minzoom is not None
    assert info.maxzoom is not None

# now
import morecantile
from rio_tiler.io import Reader

tms = morecantile.tms.get("WebMercatorQuad")
with Reader("cog.tif") as src:
    _ = src.tile_exists(0, 0, 0, tms=tms)
    _ = src.tile(0, 0, 0, tms=tms)
    _, _ = src.get_zooms(tms)
    _ = src.geographic_bounds(tms.rasterio_geographic_crs)
    
    info = src.info()
    assert info.bounds == src.bounds == src.dataset.bounds
    assert info.crs == src.crs == src.dataset.crs
    assert not hasattr(info, "minzoom")
    assert not hasattr(info, "maxzoom")

[AndrewAnnex]

I think the default CRS here should not be WGS84 but the CRS returned from self.crs.geodetic_crs As a asset probably should have a CRS, the geographic crs here should be appropriate to the CRS of the dataset itself.

[vincentsarago]

@AndrewAnnex the problem is that rasterio CRS class doesn't have geodetic_crs 😬

[AndrewAnnex]

ahh well that would be a problem. But I have an idea and will post in the new issue

[AndrewAnnex]

hmm why not opt instead to use a urn/uri rather than attempt a epsg conversion here? those should work with the other authorities and be relatively compact. Failing through to wkt makes sense but also think wkt2_2019 should be preferred as it's more verbose and complete.

[vincentsarago]

I feel I went too far with this PR (thanks for a good night of 🛏️)

I've taken another approach over #746