Broken 'reconnect' usage between tmt run invocations

[lukaszachy]

It should possible to keep the rundir and use already provisioned machine in subsequent tmt run commands. But doing that hits the traceback for 'ssh' guests.

Reproducer:

$ cat plan.fmf
execute:
    how: tmt
    script: echo

$ tmt run --id BUG provision -h virtual
...
summary: 1 guest provisioned

$ tmt run --id BUG discover prepare
plan failed

The exception was caused by 1 earlier exceptions

Cause number 1:

    prepare step failed

    The exception was caused by 1 earlier exceptions

    Cause number 1:

Full traceback (TMT_SHOW_TRACEBACK=full) shows that assert fails:

            File /usr/lib/python3.13/site-packages/tmt/steps/provision/__init__.py, line 1469, in _ssh_master_socket_path
              assert isinstance(self.parent, tmt.steps.provision.Provision)

              self = <tmt.steps.provision.testcloud.GuestTestcloud object at 0x7f7a26529590>

tmt/tmt/steps/provision/__init__.py

Line 1469 in cc75eb6

assert isinstance(self.parent, tmt.steps.provision.Provision)

As in the tmt run --id BUG discover prepare the ssh connection is established during the prepare and thus the self.parent is tmt.steps.prepare.Prepare making the assert fail.

[psss]

It would be good to get this fixed soon, proposing for 1.40.