rework deployment configuration (using less of the express mechanism) so that configuration options are available at all levels

rework deployment configuration (using less of the express mechanism) so that configuration options are available at all levels

break out substitution framework from top level run harness.  this will be used to perform hostname substitution in dev and prod environments.

(test server) fix content-length: the number of bytes, not the number of chars (fun with node.js buffers!).

fix test harness.  you really must jump through hoops to create a proper substition output filter in connnect.

don't reference external resources from browserid.org to not piss off SSL, and also a small pile of fixes to the testing harness -- specifically its handling of binary data, etc.

update testing harness to be compatible with new browserid run code.  host substitution is done at a much lower level to leverage more of the connect framework

simplify verifier a bit to make it simpler to use either off the same domain as browserid.org OR on its own domain.

more substitution fun in test harness.  find places where hostname:port is explicit without scheme prepended and get those subbed correctly (like in host-meta files)

in source urls are now all (interim == eyedee.me) live, the test harness subs in testing urls & hostnames

refactor code a bit to move connect setup into run.js of the specific server.  this will make it easier to figure out how to run servers both standalone and in the test harness

migrate to the connect framework for node.js, specifically this provides session management middleware

add support to top level runner to support server specific dispatch code (so authority can have it's own little server, for instance)

hacked up node.js test harness to run everyone on their own host/port and  to sub in ip/port combinations for hostnames.