move sigma_los^2 caching from likelihood.py to disperion.py

[achiappo]

The current instructions to cache the value of sigma_los^2 in likelihood.py (ll 23-33) are not enabled, even because caching at this stage is pointless, since every parameter changes value at this point. I, therefore, suggest moving these instructions to dispersion.py module. I would do the following:
create two separate cache containers, one for integral (the integral of M(s/r0) nu(s/rh) Ker(s,ra-beta), so where most running variables enter) and one for the surface brightness. The latter, in particular, never changes when the stellar density parameters are fixed.
In this way all three terms (integral , self.stellar.surface_brightness(R) , Jred) are treated in the same way and all values are cached