No change mdb (Exomol, HITRAN, Hitemp) from opa

address the side effect raised in #510

The following is a simple test of the side effect:

tests/unittest/opa/sideeffect_test.py

def test_sideeffect_call():
    nus, wav, res = wavenumber_grid(
        22920.0, 23100.0, 20000, unit="AA", xsmode="premodit"
    )
    mdb = mock_mdbHitemp(multi_isotope=True)
    opa1 = OpaPremodit(mdb, nu_grid=nus, allow_32bit=True, auto_trange=[500.0, 1000])
    opa1_orig = copy.deepcopy(opa1)
    opa2 = OpaPremodit(
        mdb, nu_grid=nus, allow_32bit=True, auto_trange=[500.0, 1200]
    )  # used the same mdb used in opa1
    print(opa1 == opa1_orig) -> True

Before this PR (i.e. recent develop branch), opa1 == opa1_orig gives False. This means the second call of OpaPremodit using the same mdb changes opa1.mdb. In this PR, based on the strategy described in #510, we avoid the side effect, i.e. opa does not change the internal state of mdb.

but, we need some modifications:

removed the internal state mdb.Tref
line_strength_ref_original was introduced for line strength at T=Tref_original
self.line_strength_ref (which depends on self.Tref) -> self.line_strength(Tref)
qr_interp(T) -> qr_interp(T,Tref) in MdbExomol
qr_interp(iso,T) -> qr_interp(iso,T,Tref) in MdbHitemp/Hitran
qr_interp_line(T) -> qr_interp_line(T,Tref) in MdbHitemp/Hitran
The deprecated attribute Sij0 was removed.

HajimeKawahara / exojax

No change mdb (Exomol, HITRAN, Hitemp) from opa #515

Unit Test Results