fix(solver): preserve exact π in periodic transcendental solve

[s-celles]

symbolic_solve(cos(x) ~ 1//2, x) returned
1.0471975511965976 + 6.283185307179586·n — a floating-point approximation — instead of π/3 + 2π·n. The defect had three collaborating sites in the periodic-equation path:

• isolate applied the left-inverse operator eagerly (acos(1//2) ⇒ Float64) instead of keeping a symbolic term that convert_consts could canonicalise downstream. Fixed by wrapping as Symbolics.term(invop, sol; type = Real) on both the periodic and non-periodic branches.

• fundamental_period returned 2pi / 1pi — Julia promotes Int * Irrational{:π} to Float64, so the period coefficient was 6.2832… / 3.1416…. Fixed by returning Symbolics.term(*, 2, pi) and Symbolics.term(*, 1, pi) for the sin/cos/csc/sec and tan/cot families. The degree- and cycle-based variants (sind, cosd, tand, cotd, cospi) are left unchanged — their periods are rationals.

• _postprocess_root rebuilt expression trees via SymbolicUtils.maketerm, which folds known-function calls on constant operands — so maketerm(BasicSymbolic, /, [π, 3], nothing) collapsed back to 1.0471… on the second iteration of the postprocess_root fixed-point loop. Fixed by rebuilding via Symbolics.term when any argument is an irrational math constant (π or e).

• The canonical-value lookup table in postprocess.jl was missing 3π/4 (acos(-√2/2)); added.

• The complex-roots angular factor in isolate for x^n ~ y used plain pi which promoted to Float64; wrapped symbolically.

Tests: new @testset "Exact transcendental solve (#1842)" in test/solver.jl covers every canonical sin/cos/tan right-hand side, the symbolic period coefficient, and non-canonical preservation (152 assertions). The pre-existing @test_broken at test/solver.jl:632 (composed-argument sin+cos case) still fails after this fix — it exercises the attract path which is out of scope here; a comment now documents that.

Docs: new example in docs/src/manual/solver.md.

Helps #1842.

[s-celles]

(@v1.12) pkg> activate .
Activating project at `~/work/tries/Symbolics.jl`

julia> using Symbolics

julia> @variables x
1-element Vector{Num}:
x

julia> Symbolics.symbolic_solve(cos(x) ~ 1//2, x)
[ Info: var"##283" ϵ Ζ
1-element Vector{SymbolicUtils.BasicSymbolicImpl.var"typeof(BasicSymbolicImpl)"{SymReal}}:
π / 3 + var"##283"*(2*π)

[s-celles]

Be aware that $-\frac{\pi}{3} + k 2 \pi$ solutions are (still) missing here.