ISF: bug in option cn_gammablk=vel_stab

[clem]

For ISF when using option cn_gammablk=vel_stab, one must use the variable qoce in the subroutine isfcav_gammats but it is only calculated afterwards in isfcav_mlt. So, one must initialize qoce before its use in isfcav_gammats.
In addition the initialization of qoce_b is probablement not the right one. It is written as the sensible flux while it should be the latent flux (see isfcav_mlt routine).

Fix:
change this line:

      IF (TRIM(cn_gammablk) == 'vel_stab' ) zqoce_b (:,:) = ptsc(:,:,jp_tem) * rho0_rcp ! last time step total heat fluxes (to speed up convergence)

by this one:

      IF ( TRIM(cn_gammablk) == 'vel_stab' ) THEN
         zqoce(:,:) = -pqfwf(:,:) * rLfusisf ! 
         zqh_b(:,:) =  ptsc(:,:,jp_tem) * rho0_rcp ! last time step total heat fluxes (to speed up convergence)
      ENDIF

And make the convergence criterion on qoce+qhc qnd not only qoce:

            zerr = MAXVAL(ABS(zqhc(:,:)+zqoce(:,:) - zqh_b(:,:)))

I'll do the changes and commit them because this routine just does not pass debug right now but it needs to be checked out by people who know (I guess Pierre)

[clem]

In 15082:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found

[clem]

In 15084:

Error: Failed to load processor CommitTicketReference

No macro or processor named 'CommitTicketReference' found