Fix depreciation for AutoReverseDiff. (#638)

* Update DynamicPPLReverseDiffExt.jl

* Update Project.toml

* Update ext/DynamicPPLReverseDiffExt.jl

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

* Update DynamicPPLReverseDiffExt.jl

* Update Project.toml

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

Author: yebai

Project.toml

@@ -41,7 +41,7 @@ DynamicPPLReverseDiffExt = ["ReverseDiff"]
 DynamicPPLZygoteRulesExt = ["ZygoteRules"]
 
 [compat]
-ADTypes = "0.2, 1"
+ADTypes = "1"
 AbstractMCMC = "5"
 AbstractPPL = "0.8.4"
 Accessors = "0.1"

ext/DynamicPPLReverseDiffExt.jl

@@ -9,12 +9,12 @@ else
 end
 
 function LogDensityProblemsAD.ADgradient(
-    ad::ADTypes.AutoReverseDiff, ℓ::DynamicPPL.LogDensityFunction
-)
+    ad::ADTypes.AutoReverseDiff{Tcompile}, ℓ::DynamicPPL.LogDensityFunction
+) where {Tcompile}
     return LogDensityProblemsAD.ADgradient(
         Val(:ReverseDiff),
         ℓ;
-        compile=Val(ad.compile),
+        compile=Val(Tcompile),
         # `getparams` can return `Vector{Real}`, in which case, `ReverseDiff` will initialize the gradients to Integer 0
         # because at https://github.com/JuliaDiff/ReverseDiff.jl/blob/c982cde5494fc166965a9d04691f390d9e3073fd/src/tracked.jl#L473
         # `zero(D)` will return 0 when D is Real.