Tempo2-style red noise model

[abhisrkckl]

The red noise model currently used in PINT is a pure power law, where the spectrum is
P(f) ∝ A**2 * f ** (-gamma)

On the other hand, the red noise model used by TEMPO2 has a spectrum of the form
P(f) ∝ A ** 2 * (1 + (f/fc) ** 2)**(-gamma/2)
Here, fc is the corner frequency (TNRedCorner).

There has been at least one paper comparing these models (https://doi.org/10.1093/mnras/staa2081).

Here is a tempo2 issue discussing this:
https://bitbucket.org/psrsoft/tempo2/issues/80/tempo2-uses-tnredcorner-even-if-not-set

It seems that tempo2 uses a corner power law spectrum whereas ENTERPRISE and TempoNest use a pure power law model.

It will be nice to have a way to use the tempo2-style model in PINT.
Maybe we can have a TNREDSPEC parameter with two options PL (power law) and CPL (corner power law), which will switch the spectral model with a sensible default (whatever that may be). Or we can distinguish the models by the presence/absence of the TNRedCorner parameter.