Fix metadata forms filter

[ezraporter]

Closes #431

This PR fixes the issue with redcap_metadata_read returning all forms when specific forms are requested in the forms parameter. In doing this I also noticed that the same bug applies to the fields parameter so I implemented the fix there as well.

My solution was to add a small utility function to redcap-metadata-read.R that converts a vector into the right format, a list of key-value pairs to append to the request body.

[wibeasley]

Thanks for doing this, @ezraporter. I was tweaking some of the test projects last night, so there might be some failures unrelated to your changes. I'll take care of those, if so.

[ezraporter]

No problem! I got R CMD check to pass locally so we may be in good shape but you never know :)

[wibeasley]

Looks good, @ezraporter. I'll tweak some things like adding checkmate statements and unit tests.

1. Will you tell me your ORCiD (eg, Stephan's) for the description file?

2. What do you mean by these comments? Isn't it ok if someone wants to pass either the fields or the forms, but not both?

   # append forms and fields arrays in format expected by REDCap API
   # If either is NULL nothing will be appended
   post_body <- c(post_body, fields_array, forms_array)

[ezraporter]

1. Will you tell me your ORCiD (eg, Stephan's) for the description file?

This is me: https://orcid.org/0000-0002-4690-8343

2. What do you mean by these comments? Isn't it ok if someone wants to pass either the fields or the forms, but not both?

   # append forms and fields arrays in format expected by REDCap API
   # If either is NULL nothing will be appended
   post_body <- c(post_body, fields_array, forms_array)

Ah that is a poorly worded comment. The behavior is what you described. Someone can pass either one and the right thing will get into post_body. I was just reminding myself that c(some_object, NULL) yields c(some_object) so this behaves correctly if, for examples, fields_array is NULL.