Fix metop reader

[mraspaud]

Add python 3 support to the eps l1b reader and provide bounding box reading.

• Closes #xxxx
• Tests added
• Tests passed
• Passes git diff origin/master **/*py | flake8 --diff
• Fully documented

[coveralls]

Coverage decreased (-0.1%) to 72.966% when pulling b6cfd9c on fix-metop-reader into 4e1e149 on master.

[codecov]

Codecov Report

Merging #441 into master will decrease coverage by 0.11%.
The diff coverage is 9.52%.

@@            Coverage Diff             @@
##           master     #441      +/-   ##
==========================================
- Coverage   73.07%   72.96%   -0.12%     
==========================================
  Files         134      134              
  Lines       17662    17689      +27     
==========================================
  Hits        12907    12907              
- Misses       4755     4782      +27

Impacted Files	Coverage Δ
satpy/readers/xmlformat.py	17.24% <0%> (ø)	⬆️
satpy/readers/eps_l1b.py	15.85% <9.63%> (-1.96%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 4e1e149...b6cfd9c. Read the comment docs.

[djhoese]

Too bad there isn't a simple solution for the memmap stuff, but otherwise looks good based on the little I know about this reader. When is get_bounding_box called? What is it used for?

[djhoese]

Also the loss of coverage isn't great.

[mraspaud]

get_bounding_box is called when instantiating the Scene when the area filter parameter is provided.

About coverage loss, yes it isn't good, but I would like to have real data to test this reader, but since we don't have a way to store and share it to travis effectively, I'd like to pass on that.

[djhoese]

Sounds good. I think you could still mock the numpy functions loading the file, but ok.

[stickler-ci]

F812 list comprehension redefines 'the_type' from line 95