Pure heating/cooling does not work as expected

[n-claes]

Issue description

PR #130 introduced decoupling of the heating and cooling terms. However, there are inconsistencies when the user supplies custom heating and/or cooling functions.

Assuming a user-specified cooling function $\Lambda(T)$ and heating function $H_0$ is always supplied (and cooling/heating is controlled through the parfile), then the following gives an overview:

cooling	heating	forced thermal balance	what happens	what should happen	correct?
no	no	-	user-defined heating + cooling functions are set, but non-adiabatic matrix elements are not added. Spectrum is still adiabatic and correct.	heating + cooling functions should return zero.	❌
yes	yes	yes	behaviour as expected.	-	✅
yes	yes	no	behaviour as expected.	-	✅
yes	no	yes/no	custom heating function is set and returns non-zero.	heating function should return zero.	❌
no	yes	yes/no	heating function set but non-adiabatic elements are not added, spectrum is adiabatic	custom heating function should be used, system should be non-adiabatic, no thermal balance	❌

Bug report

Minimal example for reproduction

module procedure user_defined_eq
    call settings%grid%set_geometry("Cartesian")
    call settings%grid%set_grid_boundaries(0.0_dp, 1.0_dp)
    k2 = 0.0_dp
    k3 = 1.0_dp

    call background%set_density_funcs(rho0_func=rho0)
    call background%set_temperature_funcs(T0_func=T0)
    call background%set_magnetic_3_funcs(B03_func=B03)

    call physics%set_heating_funcs(H_func=H0)
end procedure adiabatic_homo_eq

real(dp) function rho0()
    rho0 = 1.0_dp
end function rho0

real(dp) function T0()
    T0 = 1.0_dp
end function T0

real(dp) function B03()
    B03 = 1.0_dp
end function B03

real(dp) function H0()
    H0 = 2.0_dp
end function H0

Version info

• Legolas version: 2.0.3
• Pylbo version: 2.0.1
• Python version: 3.11

[n-claes]

Traced this down to the fact that users can still set custom physics functions in mod_physics, while Legolas checks for enabled physics when adding the matrix elements in mod_matrix_manager -- element blocks corresponding to disabled physics are simply skipped.

However, in the case of heating/cooling we only check if cooling is enabled, regardless of heating, which is also contained in mod_heatloss. Should be an easy fix:

• do not allow setting custom functions if the corresponding settings instance is not enabled.
• add an additional check for enabled heating when adding heatloss matrix elements.
• ❗ add unit tests for these specific cases.