feat: spell the wedge product ^, as Python does

[wardjm]

Closes #14.

wedge(α, β) was the only spelling of the wedge, while * on two Forms is the
tensor product — so Python code ported by eye silently produced a Tensor02
instead of a Form, with no MethodError to catch it.

• Base.:^(a::Form, b::Form) = wedge(a, b), so Python's α ^ β transfers directly.
• wedge (and ^) now accept a ScalarFunction on either side, where the degree-0
  form makes the wedge degenerate to the pointwise product, as in Python.
• README documents the * divergence in the field-types snippet and in Conventions.

Two deliberate non-choices, both surfaced by /code-review:

• No ^ on two ScalarFunctions. Python falls back to multiplication there, but
  f^2 already means the pointwise power (Base.:^(::ScalarFunction, ::Number)),
  so f^h meaning the product would make one operator mean two unrelated things on
  one type — a silent trap of the same species this issue exists to kill. f * g is
  the spelling for that product; a test pins wedge(f, h) as a MethodError.
• The precedence note is about +, not *. Python's ^ binds looser than
  addition, Julia's binds tighter: α ^ β + γ is α ∧ (β + γ) in Python but
  (α ∧ β) + γ here. Both are valid forms of the same degree, so it is a silent
  wrong answer — the README says so and says to parenthesise when porting.

Full suite green (2396 tests).