Multi band for MODIS

[Kirill888]

Reason for this pull request

To support modis collection 6 (MCD43A1) data files, or more generically HDF files that have multi-band datasets in them.

Proposed changes

• Add extra parameter band in image/bands/{band,path,layer}.
  • When band parameter is present and contains an integer, read only that band.
  • Original behaviour when layer is used as band indicator still present
• Updated modisprepare script and modis product definition
  • BRDF_Albedo_Parameters_xxx is now 3 layers BRDF_Albedo_Parameters_xxx_{iso,vol,geo}

Not done items

• Tests added / passed
• Fully documented, including docs/about/whats_new.rst for all changes

[coveralls]

Coverage decreased (-0.03%) to 81.124% when pulling 7978295 on multi-band-for-modis into e9b65f1 on develop.

[mpaget]

Could you hold off merging this for a couple of days if practical, please? I'm updating and working through the modis c6 collections and might propose a more consistent (between modis products) approach to naming and structuring. I suspect your changes are quite valid; I'd just appreciate a couple of days to absorb them in the context of general refinements to the way modis is described and handled.

[Kirill888]

No worries Matt,

won't be merged just yet, but do have a look at what I'm proposing, both code-wise and at the product description level. It's not a huge change, and is backwards compatible as far as code goes. Product modis_mcd43a1_tile is however affected in a way that will require re-indexing.

Also somebody needs to verify that I used the right names/bands for iso,vol,geo parameters.

[woodcockr]

Making a note, as I assume this happened whilst Matt was at GA today, that Matt confirmed his review for this change was consistent per his comments 2 days ago.

Could those accepting pull requests please get into the habit of noting that outstanding issues were resolved so others in the community know. Thank you.

[mpaget]

Matt didn't make it to GA today - hopefully tomorrow. In off-git emails we agreed it was fine to proceed with the change to support landsat nbar requirements prior to Simon going on leave. We can always re-jig modis fields/labels/descriptions in the near future. In a way I was more flagging that I was actively working on the modis indexing side of things.