commitmonitor Code

Implement using an svnrobots.txt file which can set certain rules to reduce 'hammering' a server:
* disallowautodiff : if set, then automatic diffs are not allowed
* checkinterval = XX : minimum interval in minutes between checks

Show an error message box when selecting a tree item, if there was an error the last time that url was accessed.

* save the last error
* add version info

* use a better, reentrant reader/write section class
* refresh the UI in a timer interval

This is a bigger change:
* only use one instance of CUrlInfos
* pass around that one instance to wherever it's needed
* guard the data in CUrlInfos with critical sections
* make sure the data is guarded only for short times
* in the monitoring thread, use a copy of the CUrlInfos data because of the point above

Fill in the url member of the CUrlInfo object when loading it from disk.

* get rid of the boost serialization (that one crashed and I couldn't figure out why), implement our own serialization instead.
* Implement timers and threads to fetch the log data and check for updates according to the settings.
* Implement username/password to pass to the repository if necessary.

First draft of populating the tree control with the URL infos

* enable all controls in the url dialog
* don't check the url in the url dialog but do that later (not implemented yet) in the main dialog or even the check thread
* add load/save methods for our persistent data

* the url dialog only needs one CUrlInfo, not the CUrlInfos object
* use a map with the URLs as a key for the CUrlInfos instead of a vector