dia_hsb with RK3

Context

• Branches impacted: main
• Reference configuration/test case: ORCA2_ICE_PISCES_ST

I looked at the dia_hsb output with RK3 (using suggested fix in #428 (closed)) with tradmp turned off. They show either conservation issue or a diagnostic issue.

dia_hsb : last time step hsb diagnostics: at it=          992  date=
       10504
 ~~~~~~~
    Temperature drift =  -0.113302238346833       C
    Salinity    drift =   4.276053916252722E-005  PSU
    volume ssh  drift =   -336.371670095423       km^3
    volume e3t  drift =   -336.371670099574       km^3

I didn't managed to run the same configuration with MLF, so I cannot give more details.

For information, in NEMO 4.2.2, the conservation is as expected (with tradmp turned off):

dia_hsb : last time step hsb diagnostics: at it=          992  date=
       10303
 ~~~~~~~
    Temperature drift =   5.920327490585195E-015  C
    Salinity    drift =  -2.232277797610355E-015  PSU
    volume ssh  drift =  -1.296875000000000E-009  km^3
    volume e3t  drift =  -2.824218750000000E-009  km^3