update periodic kernel description in GPstuff

[avehtari]

From Lu Cheng:

"It was said in GPStuff manual page 42 that periodic kernel was coming from this paper http://jmlr.org/proceedings/papers/v33/solin14.pdf

In page 907, equation (23) and GPStuff appendix, there is the canonical periodic covariance function. And it is not obvious to find the explicit form of quasi-periodic covariance function in section 3.5.

In the demo_periodic.m, there is alway the decay term, i.e. another SE term, which is very confusing. I suggest add an additional example which set decay to 0 and not including the extra length scale."

[kaare-mikkelsen]

how about changing the description around (121) in the manual and the help-description in the function to actually match? help gpcf_periodic talks of lengthScale, lengthScale_sexp,"decay" and period. (121) deals in "inverse length of period", and "smoothness in dimension x".

Both descriptions should include the same number of parameters, preferably under the same names. It is not at all clear what the differences between lengthScale and lengthScale_sexp is, especially since (121) can be viewed as a combination of a sine and a sexp

[avehtari]

Thanks for the suggestion.