update troubleshooting document

[wibeasley]

The troubleshooting document hasn't changed much since @nutterb and I wrote it in 2014 (#23). There are a few things that should be added/updated.

• include the API playground
• verify the metadata can be retrieved before trying the data itself (would have helped diagnose redcap_metadata_read, "object 'ds' not found"  #269 & REDCapR: redcap_read function gives error in R version 3.6.1 and not 3.5.3  #272 a little faster)
• switch from RCurl to httr for vanilla-R section
• change from third-person to second-person. (e.g., "verify the user" to "verify")
• add new "Common Language-Agnostic Gotchas" section
• add new "General Troubleshooting Advice" section
• include step for validate_for_write()

@nutterb -I'm adding the metadata section for redcapAPI. I'll tell you when I'm done with this checklist, and you're more than welcome to change anything.

@vortexing & @sybandrew, you've been through this recently (with #269 & #272). Any suggestions how it could have helped your cases faster?

Everyone else -we'd like this document to stay useful to as many users & institutions as possible. Feel free to propose anything that you think will help.

• Announce update in the forums and invite others to PR.

The latest master version is available. The dev version might lag by a few days.

[vortexing]

I think my issue was mainly that I didn't realize that the problem was in my data dictionary rather than an R problem or an API problem. If the error message when the content was not able to be transformed into a data frame indicated that the problem was a data dictionary parsing problem, then I would say that the help text you have here to go look at my dictionary in something that isn't Excel would be perfect. I wonder in the help text here: https://ouhscbbmc.github.io/REDCapR/articles/TroubleshootingApiCalls.html#gotchas-agnostic would be best to include possible error messages like I posted so people would be more likely to realize this was actually their problem?