README: Indicate that code results may not be reproducible

[frankfarach]

See ropensci/software-review#505 (comment):;

The example in the README is not reproducible, meaning I got different results when I ran the code than what was shown in the README output. Also, the ouput included in the vignette differ from those in the README. I assume this is due to a change in the NPI itself between when the README was created and when I ran the code? If possible, reproducible example in the README is ideal, though I understand that may be a challenge with a changing API.