Currently, based on a long-ago understanding of what other major search engines/crawling projects did, Heritrix treats anything other than a '200' from a /robots.txt request as a robots-not-present situation.
It appears that Google, at least, follows redirects and respects the directives so found. (See for example the query [site:www.swims.nhs.uk] regarding the site mentioned in thread http://tech.groups.yahoo.com/group/archive-crawler/message/8078.)
Heritrix should match this behavior, since most webmasters craft their robots.txt based on expectations set by Google.
Open question: does Google follow redirects even to URIs that don't end "/robots.txt", or the URIs on different hosts, and still apply the resulting rules to the original hostname?