176 use wwinference draws

[kaitejohnson]

This PR closes #176. It does the following:

• Replaces the function to get a draws dataframe that was originally in wweval to use the one in wwinference::get_draws() for binding the data to the posterior estimates
• Formats the output to match what is expected by the rest of the wweval package
• Removes the dependency on cfaforecastrenewalww in wweval, the container builds, and the pipeline code
• No longer pre-compiles the model in the container build (I couldn't figure out how to do this, since we were previously specifying a relative path within this directory, and now we are using the system file in the package. Currently this works by compiling the model within each VM)

To make sure this works as expected I:

• tested locally
• tested in the command line
• tested on azure for a single location forecast date

Next steps:

• rerun all location/forecast dates for fit and post process using wwinference
• try to run the targets_eval_postprocessing.R pipeline, after #179 is also merged. I anticipate there may be a few additional modifications here since a part of this relies on the input data, some of which is now formatted slightly differently (column names differ) . I will change the functions in wweval accordingly. The scores, quantiles, etc should be formatted the way that wweval expects them downstream.

[gvegayon]

I think this is getting really close to being able to remove cfaforecastrenewalww dependency from wweval/the container build! @gvegayon Do you want me to do that in this PR or do you want to work on in a separate branch merged off of lesscfaforecastrenewalww-2?

This PR is not big, so I think we still have some space to do stuff, so I would say, "go for it, @kaitejohnson!"

[kaitejohnson]

Ok will test out trying to remove and see how it goes!