Warn users about getting readonly launchpad transports

[Aaron Bentley]

Martin Pool wrote:
> I believe the problem is that at the code level we don't know whether
> the user is going to write to it or not until after the redirection
> has been done.  (And indeed, they may be getting a checkout but later
> wanting to write to it.)
> 
>>  Anyhow, it's a step forward  :)
> 
> +1, I agree, though as Aaron recently touched this redirector it could
> be good to check with him too.

I would really rather we added a hint to "Branch.open", transport.open,
etc, so that we would know whether we were going to write at transport
open time.  Then we could fail appropriately if trying to open a
readonly transport for write operations.

> James, please put [merge] in your subject line (even if you're not
> totally sure it should merge) so that BB will track it.  I actually
> think it should track all patches; the number of things that look like
> patches but are not intended to merge is pretty negligible.

I still disagree, especially, we get patches for the stats plugin,
fastimport, looms, bzr-email and that's just going back to January.

Aaron