Derive the candidate-line Big-M from the nodal angle bound

[erikfilias]

Change

The angle side of the candidate-line Big-M is a literal pi:

M_angle_lca = (1.0 + pMBigMEpsilon) * max(par['pLineNTCBck'][lca], math.pi * par['pSBase'] / par['pLineX'][lca])

and the nodal angle bound, a few lines below in the same function, is pi/2:

par['pMaxTheta'] = par['pDemandElec']*0.0 + math.pi/2

The angle difference across a line is bounded by twice the nodal bound, so pi is valid only while
the nodal bound is pi/2. Nothing ties the two together. This names the bound once and derives the
Big-M from it.

Effect

None on results. 2.0 * (math.pi/2) is bit-identical to math.pi in IEEE-754, so the Big-M is
unchanged for every line. No interface or input-data change.

Changing pMaxTheta currently leaves the Big-M at its old value, which is then too small for the
wider angle range and restricts flows on candidate lines without raising an error. After this
change the two move together.

Scope

Making pMaxTheta configurable is not included. If added later, oT_Data_Parameter would be the
natural place, and two things would need handling: a MaxTheta column there is read by the generic
parameter loader and then overwritten by the broadcast assignment in
openTEPES_DataConfiguration.py, and the near-binding warning in openTEPES_OutputResultsNetwork.py
hardcodes math.pi / 2.