`deepcopy`ing NamedTuples seems to break

[penelopeysm]

MWE:

import Enzyme: Enzyme, Forward, Reverse, Const

nt = (a = 0.0,)
x = [0.5]

function f(x, nt)
    nt2 = deepcopy(nt)
    return nt2.a + x[1]
end

f(x, nt) # 0.5

Enzyme.gradient(Forward, f, x, Const(nt))
#=
ERROR: MethodError: no method matching deepcopy_rtact(::@NamedTuple{a::Float64}, ::@NamedTuple{a::Float64}, ::IdDict{Any, Any}, ::@NamedTuple{a::Float64})
The function `deepcopy_rtact` exists, but no method is defined for this combination of argument types.
=#

Enzyme.gradient(Reverse, f, x, Const(nt))
#=
ERROR: AssertionError: !(typeof(x) <: Active)
Stacktrace:
  [1] macro expansion
    @ ~/.julia/packages/Enzyme/dM66n/src/compiler.jl:5491 [inlined]
  [2] enzyme_call
    @ ~/.julia/packages/Enzyme/dM66n/src/compiler.jl:5025 [inlined]
  [3] AugmentedForwardThunk
    @ ~/.julia/packages/Enzyme/dM66n/src/compiler.jl:4964 [inlined]
  [4] runtime_generic_augfwd(activity::Type{…}, runtimeActivity::Val{…}, strongZero::Val{…}, width::Val{…}, ModifiedBetween::Val{…}, RT::Val{…}, f::typeof(deepcopy), df::Nothing, primal_1::@NamedTuple{…}, shadow_1_1::Base.RefValue{…})
    @ Enzyme.Compiler ~/.julia/packages/Enzyme/dM66n/src/rules/jitrules.jl:422
=#

versioninfo and stuff

(ppl) pkg> st
Status `~/ppl/Project.toml`
  [7da242da] Enzyme v0.13.48

julia> versioninfo()
Julia Version 1.11.5
Commit 760b2e5b739 (2025-04-14 06:53 UTC)
Build Info:
  Official https://julialang.org/ release
Platform Info:
  OS: macOS (arm64-apple-darwin24.0.0)
  CPU: 10 × Apple M1 Pro
  WORD_SIZE: 64
  LLVM: libLLVM-16.0.6 (ORCJIT, apple-m1)
Threads: 1 default, 0 interactive, 1 GC (on 8 virtual cores)

[wsmoses]

fixed by #2510