Fix report generation for models with quads and plates (fixes #268)

[gaoflow]

Fixes the remaining reporting breakage from #268 (the quad/plate case @boustrephon reported, which was never addressed).

Problem

Reporting.create_report() raises KeyError: 'Combo 1' for any model containing quads or plates — e.g. Examples/Shear Wall - Advanced.py:

from Pynite import Reporting
Reporting.create_report(model, format='html', output_filepath='./ModelReport.html')
# KeyError: 'Combo 1'  (in Quad3D.D -> self.i_node.DX[combo_name])

Root cause

Quad3D/Plate3D shear(), moment() and membrane() all have the signature

def shear(self, xi=0.0, eta=0.0, local=True, combo_name='Combo 1'): ...

The report template called them as quad.shear(-1, -1, combo.name) — so the combo name was bound to local, and combo_name kept its default 'Combo 1'. Since most models never analyze the default 'Combo 1', the subsequent node.DX['Combo 1'] lookup raised KeyError. (The member.* calls are unaffected — their third positional arg genuinely is combo_name.)

A second, latent template bug was masked behind the first: membrane() returns a (3, 1) column vector, but several cells indexed it as [n] instead of [n][0], leaving a 1-element array that the "%.3g" | format(...) filter cannot render (TypeError: only 0-dimensional arrays can be converted to Python scalars).

Fix

• Pass combo_name=combo.name as a keyword on every plate/quad shear/moment/membrane call (80 sites) so local keeps its intended default.
• Normalize membrane result indexing to [n][0] (24 sites), matching shear/moment and the already-correct membrane cells.

Template-only change — no solver code touched.

Verification

• Examples/Shear Wall - Advanced.py + create_report(...) now renders a complete 8 MB HTML report (previously raised immediately).
• New Testing/test_reporting.py: builds a supported quad and plate under pressure, analyzes a single custom-named combo (1.4D) so the default Combo 1 is never solved, generates the HTML report, and asserts it does not raise, that the analyzed combo appears, and that Combo 1 does not. The test fails on main (KeyError: 'Combo 1') and passes with this change.
• Beam-only reports (no quads/plates) still render unchanged.
• Full core suite: 116 passed (115 + the new test).

Note: #268 originally bundled three problems; the first two were resolved in v1.2.0 / v1.5.0, and this addresses the last outstanding one. Happy to retarget to a dedicated issue number if you'd prefer.