Qmec: new parameterization

[benRenard]

After discussion with @JeromeLeCoz and @Felipemendezrios:

Reparameterize as follows:
A0=Be*(y0-be)
and Be disappears from the parameter vector.

with A0 a "representative wetted area". y0 can be computed from the inputs as the mean mean stage, i.e. average(0.5*(h1+h2))

Naming:
Qmec0 should be the original model from Bourgault and Matte (currently: SFDTidal_Qmec_model)
Qmec should be the version we prefer, hopefully this reparameterization
Qmec2 should be Clara's version (currently: SFDTidal_Qmec2_model)
Consider removing "SFDTidal" from the name, or possibly just "Tidal" ->SFD_Qmec?

[benRenard]

Done in commit 80fd3d8, but it was slightly more complicated than expected: the proposed reparameterization still led to huge correlations between $n_e$ and $b_e$. The reparameterization finally implemented is as follows:

1. $A_0=B_e(y_0-b_e)$
2. $\phi=\frac{n_e^2}{(A_0(y_0-b_e))^c}$

$\implies$ original parameters $B_e$ and $n_e$ disappear and are replaced in Qmec equation by:

1. $B_e=A_0/(y_0-b_e)$
2. $n_e^2=\phi*(A_0(y_0-b_e))^c$

Note that $A_0$ and $\phi$ are new Qmec parameters that need to be calibrated, unlike $y_0$ which is a fixed, user-provided typical water level.

[JeromeLeCoz]

OK, but I feel like $\phi$ could take the form of the (dimensionless) Darcy-Weisbach friction coefficient $f=\frac{8gn^2}{R_h^{1/3}}$, often deemed to be better posed than the dimensional Manning or Strickler friction coefficients.

The Qmec equation with our notations writes:

$\frac{\Delta Q}{\Delta t} = -gB_e(\bar{y}-b_e)\frac{\bar{y}+\delta}{\Delta x} - gn_e^2 \left[ \frac{B_e+2(\bar{y}-b_e)}{B_e(\bar{y}-b_e)} \right]^{c} \frac{Q|Q|}{B_e(\bar{y}-b_e)} + \frac{2Q}{\bar{y}-b_e}\frac{\Delta\bar{y}}{\Delta t}$

with $c=4/3$ (could be fixed, in my opinion).

You could test the following reparameterization:

1. $A_0=B_e(y_0-b_e)$
2. $f=8gn_e^2\left[\frac{A_0/(y_0-b_e)+2(y_0-b_e)}{A_0}\right]^{c-1}$

⟹ original parameters $B_e$ and $n_e$ disappear and are replaced in Qmec equation by:

1. $B_e=A_0/(y_0-b_e)$
2. $n_e^2=\frac{f}{8g}\left[\frac{A_0}{A_0/(y_0-b_e)+2(y_0-b_e)}\right]^{c-1}$

The fixed, user-provided typical water level $y_0$ is used to define $A_0$ and to convert friction factor $n_e$ into dimensionless friction factor $f$.