ENH: signal: add type=False as parameter for periodogram and welch

[mawesi]

Add the option to give type=None as parameter to signal.detrend. This will turn off detrending and return the input data. This is especially useful for functions using signal.detrend internally, such as signal.welch or signal.periodogram.

[coveralls]

Coverage remained the same when pulling ac37880 on mawesi:detrend_none into b52aba6 on scipy:master.

[pv]

None is used usually as keyword argument placeholder, so it would be better to use False here.

[mawesi]

Ok, I see the point. I have updated to type=False.

[coveralls]

Coverage remained the same when pulling fa8f2ed on mawesi:detrend_none into b52aba6 on scipy:master.

[rgommers]

I agree with the last comment on the mailing list that this looks weird. Why have a do-nothing option in detrend? This should be in welch and periodogram imho.

[mawesi]

It will defininitely look strange for a user of detrend to have such an option. It only would make sens if called from other functions which then automatically have an option to turn detrending off. The question is if detrend has to care about these functions calling it. And I agree that this should rather be fixed in the calling function.
So we come back to the original proposal to change it in welch, periodogram and other (future) functions which call detrend internally?

[rgommers]

@mawesi yes, +1 from me

[coveralls]

Coverage remained the same when pulling 879c002 on mawesi:detrend_none into c08dd62 on scipy:master.

[rgommers]

Added a test and merged this in fd1377e. Thanks @mawesi