Clarify CPL dark-energy density scalings across CAMB and Halofit

[Copilot]

The issue was a mismatch between CPL dark-energy density expressions in DarkEnergyInterface.f90 and halofit.f90. The underlying (w0, wa) definition is unchanged; the apparent discrepancy came from different routines evolving differently rescaled density quantities.

• What changed

  • Added targeted source comments in fortran/DarkEnergyInterface.f90 and fortran/halofit.f90 to distinguish:
    • rho_de / rho_de0
    • a^2 * rho_de / rho_de0
    • a^4 * rho_de / rho_de0
  • Kept the numerical implementation unchanged; the update is a clarification of variable meaning and normalization.

• Regression coverage

  • Extended camb/tests/camb_test.py with a focused CPL background test against the public API.
  • The test checks that get_dark_energy_rho_w() matches the expected CPL density and equation-of-state relations.

• CPL relation now made explicit

  a = np.array([0.3, 0.5, 0.8, 1.0])
  expected_rho = a ** (-3 * (1 + w0 + wa)) * np.exp(-3 * wa * (1 - a))
  expected_w = w0 + wa * (1 - a)

• Net effect

  • Makes it clear that the formulas were already consistent.
  • Reduces the chance of interpreting internal helper variables as using a different CPL parameter convention.