Add symbolic + structural math validation

[Copilot]

The existing M[...] mode accepts any symbolically equivalent expression, which can wrongly mark variable-swapped answers as correct (e.g., w_C = 0 accepted for w_B = 0, or D*x + C accepted for C*x + D). This PR introduces MV[...] (V for variable-aware) to require both symbolic equivalence and structural variable-role consistency, while still accepting any valid mathematical arrangement.

How MV works:

• Symbolic simplification is applied before comparison, so simplified forms (e.g. M^2*C^2/M) are correctly accepted.
• Any commutative reordering of terms or factors is accepted (e.g. D + C*x matches C*x + D).
• Answers where variables take structurally different roles are rejected even when mathematically equivalent (e.g. D*x + C does not match C*x + D).
• For equation answers, terms may additionally be freely rearranged between the two sides or the sides may be swapped (e.g. E/m = c^2 is accepted for E = m*c^2).

[Tom-van-Woudenberg]

Tested here: https://oit.tudelft.nl/CTB2210/test_MV/krachtenmethode_vakwerk/lesoefeningen.html
(very limited testing)

[douden]

Shouldn't this be a (necessary) improvement of M, instead of its own type?

[Tom-van-Woudenberg]

@douden, if we would make this the standard behaviour of M[], do you see any issues?

[douden]

I would not see any issues.