Merge #172

172: Add args/kwargs to `optimize_hyperparameters!()` r=odunbar a=odunbar

Resolves #168 to add args + kwargs into `optimize_hyperparameters!()` for `GPJL()` optimizer
## Content
- Adds ability to modify the arguments for `Optim` methods. 
- Adds to docstring, a warning about the positional default argument for `method` forced upon us in `GaussianProcesses.jl`

## Examples 
I've verified these work in `examples/Emulator/GaussianProcess/learn_noise.jl` example. after adding `Optim,LineSearches` to `examples/Emulator/GaussianProcess/Project.toml`)

1. Add kwargs only 
```julia
...
kwargs = (key1="val1", key2="val2")
optimize_hyperparameters!(emulator, kwargs...)
```
2. Modify args only (the user MUST add `Optim` and set the first argument to be the `method`, default `LBFGS()`)
```julia
using Optim
...
other_args = (LBFGS(), arg1, arg2) 
optimize_hyperparameters!(emulator, args...)
```
3. Modify the method linesearch [option](https://julianlsolvers.github.io/LineSearches.jl/stable/) (relating to #168)
```julia
using Optim, LineSearches
...
method_arg = LBFGS(linesearch=BackTracking())
optimize_hyperparameters!(emulator, method_arg)
```



Co-authored-by: odunbar <odunbar@caltech.edu>

src/Emulator.jl

@@ -153,8 +153,8 @@ $(DocStringExtensions.TYPEDSIGNATURES)
 
 Optimizes the hyperparameters in the machine learning tool.
 """
-function optimize_hyperparameters!(emulator::Emulator{FT}) where {FT <: AbstractFloat}
-    optimize_hyperparameters!(emulator.machine_learning_tool)
+function optimize_hyperparameters!(emulator::Emulator{FT}, args...; kwargs...) where {FT <: AbstractFloat}
+    optimize_hyperparameters!(emulator.machine_learning_tool, args...; kwargs...)
 end
 
 

src/GaussianProcess.jl

@@ -168,13 +168,16 @@ end
 $(DocStringExtensions.TYPEDSIGNATURES)
 
 Optimize Gaussian process hyperparameters using in-build package method.
+
+Warning: if one uses `GPJL()` and wishes to modify positional arguments. The first positional argument must be the `Optim` method (default `LBGFS()`).
 """
-function optimize_hyperparameters!(gp::GaussianProcess{GPJL})
+function optimize_hyperparameters!(gp::GaussianProcess{GPJL}, args...; kwargs...)
     N_models = length(gp.models)
     for i in 1:N_models
         # always regress with noise_learn=false; if gp was created with noise_learn=true
         # we've already explicitly added noise to the kernel
-        optimize!(gp.models[i], noise = false)
+
+        optimize!(gp.models[i], args...; noise = false, kwargs...)
         println("optimized hyperparameters of GP: ", i)
         println(gp.models[i].kernel)
     end
@@ -270,7 +273,7 @@ function build_models!(
 end
 
 
-function optimize_hyperparameters!(gp::GaussianProcess{SKLJL})
+function optimize_hyperparameters!(gp::GaussianProcess{SKLJL}, args...; kwargs...)
     println("SKlearn, already trained. continuing...")
 end