Permit dot call (like Distributions.Normal) to be used in model definition

[sunxd3]

The motivation for the PR is initially to allow the use product_distribution in JuliaBUGS model. By writing

@bugs begin
    x[1:2] ~ Distributions.product_distribution(fill(Normal(), 2))
end

It also enables

julia> foo(x) = x + 1
foo (generic function with 1 method)

julia> model_def = @bugs begin
           a = Main.foo(b)
       end
quote
    a = Main.foo(b)
end

julia> compile(model_def, (;b=2))
BUGSModel (parameters are in transformed (unconstrained) space, with dimension 0):

  Model parameters:

  Variable sizes and types:
    a: type = Int64
    b: type = Int64

i.e., an easier way to introduce external functions into JuliaBUGS.

This can't fully replace @register_primitive yet, because using Main module is relying on Julia runtime behavior and not very intuitive. Examples:

julia> module TestModule
           bar(x) = x + 1
       end
Main.TestModule

julia> model_def = @bugs begin
           a = TestModule.bar(b)
       end
quote
    a = TestModule.bar(b)
end

julia> compile(model_def, (; b = 1))
ERROR: UndefVarError: `TestModule` not defined in `JuliaBUGS`
...

one would need to do

julia> model_def = @bugs begin
           a = Main.TestModule.bar(b)
       end
quote
    a = Main.TestModule.bar(b)
end

also

julia> @testset "t" begin
           foo1(x) = x + 1
           model_def = @bugs begin
               a = Main.foo1(b)
           end
           compile(model_def, (; b= 1))
       end
t: Error During Test at REPL[18]:1
  Got exception outside of a @test
  UndefVarError: `foo1` not defined in `Main`
...

[coveralls]

Pull Request Test Coverage Report for Build 11836332251

Details

• 2 of 2 (100.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.08%) to 74.173%

---

Totals
Change from base Build 11832739597:	0.08%
Covered Lines:	1390
Relevant Lines:	1874

---

💛 - Coveralls

[github-actions]

BridgeStan not found at location specified by $BRIDGESTAN environment variable, downloading version 2.5.0 to /home/runner/.bridgestan/bridgestan-2.5.0
Done!
Model dogs produces error: ErrorException("log_density() failed with exception: Exception: bernoulli_lpmf: Probability parameter is inf, but must be in the interval [0, 1] (in '/home/runner/work/JuliaBUGS.jl/JuliaBUGS.jl/benchmark/stan/bugs_examples/vol1/dogs/dogs.stan', line 37, column 6 to line 38, column 62)\n")

Model	Parameter Count	Data Count	Stan Density Time (µs)	Stan Density Gradient Time (µs)	JuliaBUGS Density Time with Graph Walk (µs)	JuliaBUGS Density Gradient Time with ReverseDiff.jl(compiled tape) (µs)
rats	65	150	5.4422	8.00833	68.829	89.948
pumps	12	10	0.994276	1.30882	11.301	6.367
dogs	2	720	NA	NA	177.276	153.277
seeds	26	21	2.53382	3.137	27.291	19.607
surgical_realistic	14	12	1.25452	1.68539	14.838	8.42267
magnesium	108	96	10.74	12.383	119.914	76.283
salm	22	18	2.42875	3.24056	20.538	12.343
equiv	15	20	2.53291	3.47525	18.675	15.269
dyes	9	30	1.00914	1.33523	11.897	13.5905
stacks	6	21	1.1822	1.78019	22.683	14.427
epil	303	236	34.334	39.9945	278.481	248.61
blockers	47	44	3.23378	3.68688	56.636	30.026
oxford	244	240	16.581	19.055	368.595	172.102
lsat	1006	5000	177.718	215.423	1858.4	1208.69
bones	33	422	73.147	91.05	410.944	179.731
mice	20	65	7.256	9.304	28.5835	38.933
kidney	64	58	10.745	16.34	61.404	70.842
leuk	18	714	20.488	26.721	212.678	201.907
leukfr	40	714	23.614	31.659	225.923	245.866