`@def name … end true` (trace mode) prints the parsed model twice

[ocots]

Context

Found while making OptimalControl's docs/src/modelling/abstract-syntax.md executable
(control-toolbox/OptimalControl.jl#913). The page has a !!! hint block demonstrating
trace mode; once executed, its output shows the parsed model twice.

Reproduce

using OptimalControl   # CTParser 0.9.4-beta, CTBase 0.30.2-beta, CTModels 0.19.2-beta

c(t) = 1.0

@def damped_integrator begin
    tf ∈ R, variable
    t ∈ [0, tf], time
    x = (q, v) ∈ R², state
    u ∈ R, control
    q̇ = v(t)
    v̇ = u(t) - c(t)
    ẋ(t) == [q̇, v̇]
    ∫(u(t)^2) → min
end true

Output (abridged — the whole block is printed twice, identical apart from gensym numbers):

variable: tf, dim: 1
time: t, initial time: 0, final time: var"tf##4005"[1]
state: x, dim: 2
control: u, dim: 1
alias: q̇ = (x[2])(t)
alias: v̇ = (var"u##4006"[1])(t) - c(t)
dynamics: ∂(x)(t) == [(x[2])(t), (var"u##4006"[1])(t) - c(t)]
objective (Lagrange): ∫((var"u##4006"[1])(t) ^ 2) → min
variable: tf, dim: 1
time: t, initial time: 0, final time: var"tf##4025"[1]
state: x, dim: 2
control: u, dim: 1
alias: q̇ = (x[2])(t)
alias: v̇ = (var"u##4029"[1])(t) - c(t)
dynamics: ∂(x)(t) == [(x[2])(t), (var"u##4029"[1])(t) - c(t)]
objective (Lagrange): ∫((var"u##4029"[1])(t) ^ 2) → min

Expected

The trace is printed once.

Impact

Cosmetic, but noisy in the docs and in interactive use — a reader could think the double
output signals a problem with their definition.

[ocots]

Fix in #347 (sibling of #345, targeting chore/deps-examodels-0.12).

Root cause: with the :exa backend active (the default), def_fun parses the definition a second time via def_exa(e; log=log) to embed the ExaModels builder, and that pass re-emitted the trace. The :exa sub-parse now runs with log=false.

Verified: the objective (Lagrange) / state: … markers go from 2 occurrences to 1.