Bugfix

[vmenezesmoura]

Little adjusments in 2 lines that were crashing tide analysis end predictions

[mdunphy]

Thanks for the PR! I cherry-picked the first and third commits to master. For the second, the ii fix is also merged via #38.

For the pearg fix, Rich's matlab version has:

  if size(inf.amprat,2)==1    % For real time series
    pearg=     2*pi*(vu(mu+ii)-vu(jref(ii))+inf.ph(ii))/360;
    pcfac=inf.amprat(ii).*f(mu+ii)./f(jref(ii)).*exp(i*pearg);
    pcorr=1+pcfac.*scarg;
    mcfac=conj(pcfac);
    mcorr=conj(pcorr);

... that is, no ii multiplying the 2*pi. I'm not sure if the inference code here is well tested; it would be prudent to have a test case from Rich's version to see if we can reproduce it here