Review of TS-798 please

[Leif Hedstrom]

Hi all,

I need at least a couple of people to review the second patch on 
https://issues.apache.org/jira/browse/TS-798 before I commit it for 
v3.0. The main issue here is that in the case of a remap.config "parse" 
error (of almost any kind), we end up adding the defunct mapping to the 
map rules, and then we free that memory. So it's a double whammy.

This code fixes that, and also cleans up the error handling a little 
bit, to be more consistent (amc: yes, I kept the goto's, I don't want to 
make more changes than necessary).

Please comment on the bug after review / testing. I would like to get 
this into the 3.0 release, since it fixes a fairly serious problem IMO.

Thanks!

-- Leif