Enable dry run mode

[shoenig]

In some cases it would be nice to see what taggit will produce without actually creating it.

For example, a CI pipeline that injects the version string at build time would need to know the next version (i.e. run taggit in -dry mode), commit the change, then run taggit again for real to create and publish the version.

Noticed I was really sloppy with the existing flags where it just gets pulled from the FlagSet. Ideally we should follow the example, which creates a local variable and passes it in by reference to be set during Parse.