Additional options for continuum fit

[marijana777]

Adding options to:

• define input and output files with knots in lt_continuumfit
• define continuum knots automatically, by including only pixels that do not correspond to absorption lines (code: analysis/continuumfnd.py )

[coveralls]

Coverage increased (+0.2%) to 71.544% when pulling a449847 on marijana777:continuumfind into f953c29 on linetools:master.

[profxj]

Looks ok overall but the code needs to be more Pythonic.

[marijana777]

Addressed all requested changes for now.

[profxj]

Thanks for this @marijana777 .
The only thing left is to add a test or two of your contknots method.
Add it to the tests folder in analysis.

[marijana777]

Thank you @profxj for the comments. Just added the test of the contknots method

[profxj]

This all looks good now @marijana777
But wait until PR #512 is merged into master
and then pull that in to yours so that Travis tests
should pass. Once they do, I'll approve and merge.

[profxj]

Ok, @marijana777 pull in master and confirm that Travis passes.
THanks.

[marijana777]

@profxj confirming that Travis passed on test_continuumfnd.py in all cases. I also confirmed that the rest of the Travis output seems almost the same as in the most recent merged PR at this moment (i.e., the same tests failed, etc).
Should we merge this PR now?