Migrate `get_job()` (singular) to v1 API

[toph-allen]

Similarly to #340, the singular get_job() uses an unversioned API and should move to the v1 endpoint.

The unversioned endpoint currently in use returns the stdout and stderr of the process. This is the only reason I imagine a user would want to use this function over the get_jobs() function.

The v1 endpoints for getting a single job just returns the same data as get_jobs(). There are two additional endpoints, though, that return the logs and any error message for a given process.

So get_job_logs() and get_job_error()

Perhaps return the logs as a data frame, with each message on a row, and separate columns for timestamp, data, and source (stdout, stderr).

Still run into the problem of having to pass in both a content item and a job key with the current approach, e.g.:

CONTENT_GUID <- "guid"

client <- connect()
item <- content_item(CONTENT_GUID)
jobs <- get_jobs(item)
key <- jobs$key[1]
logs <- get_job_logs(item, key)

If get_jobs() returned a list of job class objects, we could just pass in the single job (the job object could contain a reference to its parent content item). But in the current mode of returning job data as a data frame, you get to, well, interact with that data with data frame semantics, which is really nice and feels like second nature to me (e.g. dplyr::filter etc.).

[toph-allen]

This will fix #296