Smarter simplification and lambdification

[mmikhasenko]

Played with the repository today, want to drop feedback here.

My goal was to get the unpolarized intensity as a function of the Mandelstam invariant variables, $\mathcal{I}(\sigma_1,\sigma_2)$.
Once you put all expressions together, BWs, angles, and couplings under the spin some $|\mathcal{M}|^2$, it take forever to doit + simplify.
I waited an hour and it did not finish, even with 2 isobars. So, likely one needs to be smart for the simplification step.
Something along this lines

1. start with parametric expressions for BW
2. simplify angular dependence, getting a bilinear form on H_prod BW H_decay
3. substitute the angles
4. substitute BWs
5. simplify
6. compile

See also ComPWA/report#7