Improve citation process error reporting

[vincerubinetti]

Currently, the citation process has a boolean error flag for reporting the errors at the end:

# exit at end, so user can see all errors in one run
if error:
    log("Error(s) occurred above", level="ERROR")
    exit(1)
else:
    log("All done!", level="SUCCESS")

We should change this to a list of strings to at least report the ids of which sources failed. In cases like greenelab.com, where there are 100s of citations, its actually quite cumbersome to scroll up to find which ids failed, especially since we're telling Manubot to be verbose in its error logging.

Could also report how many errors vs warnings were encountered.

Also, in the log utility function, prefix the messages with their level and/or icon, i.e. "⚠️ WARNING: couldn't parse manubot response". Currently, the level is only indicated by color, which could fail to render properly in certain environments. And, if you use the search feature in GitHub Actions workflow logs, it messes with the color because it tries to highlight the matches. Also, having the words "error" and "warning" would make it easier to ctrl+f through the logs.