Added fast processing for RINEX 3

[breid-phys]

The RINEX 3 code was re-written following the general idea behind the fast RINEX 2 code, loading all of the values into a numpy array before converting to xarray at the end.

The new code is significantly faster than before, and passes all of the automated tests. If the 'fast' flag is false, it should use the same technique as before.

All of the changes should be within rinexobs3().

[scivision]

Thank you this has been a long-requested feature!

[VOMAY]

I tested this today and it works well. I highly recommend to merge this!

This will close: #23

[scivision]

Thanks, I had to fix other bugs first related to xarray changes. Thanks for your changes and endorsement!

[ielenik]

Looks like the line [186] is incorrect
isv = [i for i,s in enumerate(svl) if s in gsv]

usually it works - if you have sats [g01,g02,g03] and this sats appears in same order, you will get correct indexes [0,1,2]
But if g01 and g03 was visible from beginning but g02 appears just now (under index 123) you will get [0,1,123] instead of correct [0,123,1]. I do not know how elegantly correct this, but using map like {'g02':123,'g01':0,'g03':2} fixes issue

[breid-phys]

Looks like the line [186] is incorrect
isv = [i for i,s in enumerate(svl) if s in gsv]

I think you are right, in the first version of the code I had made an indexing mistake. I had to re-write the fast processing again since it wasn't fast enough, it is a lot more elegant now. I don't think the line you are referring to exists in the code anymore, though.

Please let me know if the problem still exists!

[ielenik]

Oh, this definitely possible. I am not a big specialist in git so that is really possible that I took some outdated version.
Sorry to bother you

[aclel]

Thanks @breid-phys, this is great. @scivision is there a plan to merge this in?