I did consider this. Facts are not copyrightable, but unique selections of facts are copyrightable. This is a basic tenet of the copyright decisions regarding databases and phonebooks.
Moreover, even if the copyright does not rise to the level of a copyright violation (if the facts copied are not sufficiently unique -- that would require a look at the CL fields, the 3Tap API, and Padmapper's use), it would still be a TOS violation, which courts have held are enforceable especially against commercial entities.
1) Databases are protected by copyright, but there is no such thing as "database rights." As I have repeatedly pointed out, the selection, arrangement, and presentation of facts is copyrightable in the U.S. if such activities involve any actual choice.
2) A copy of the Craiglist site is visited. The fact that the copy is hosted by the Google cache is immaterial for copyright or TOS purposes, unless Google cache did not have any right to keep that copy. Google operates the cache under the fair use doctrine, and so has compulsory rights to the Craiglist website. Ergo, visiting the cached copy is the same as visiting the original site, for legal purposes. If it were not, then all TOS could easily be avoided simply by accessing a CloudFront or Akamai-hosted website, or through any third-party aggregation service or software.
Moreover, even if the copyright does not rise to the level of a copyright violation (if the facts copied are not sufficiently unique -- that would require a look at the CL fields, the 3Tap API, and Padmapper's use), it would still be a TOS violation, which courts have held are enforceable especially against commercial entities.