Handle cancelled-before-proc-created spawn case

It's definitely possible to have a nursery spawn task be cancelled
before a `trio.Process` handle is ever returned; we now handle this
case as a cancelled-during-spawn scenario. Zombie collection logic
also is bypassed in this case.

tractor/_spawn.py

@@ -230,6 +230,8 @@ async def new_proc(
             ]
 
         cancelled_during_spawn: bool = False
+        proc: Optional[trio.Process] = None
+        try:
             try:
                 proc = await trio.open_process(spawn_cmd)
 
@@ -238,9 +240,9 @@ async def new_proc(
                 # wait for actor to spawn and connect back to us
                 # channel should have handshake completed by the
                 # local actor by the time we get a ref to it
-            try:
                 event, chan = await actor_nursery._actor.wait_for_peer(
                     subactor.uid)
+
             except trio.Cancelled:
                 cancelled_during_spawn = True
                 # we may cancel before the child connects back in which
@@ -320,6 +322,7 @@ async def new_proc(
             # killing the process too early.
             log.cancel(f'Hard reap sequence starting for {uid}')
 
+            if proc:
                 with trio.CancelScope(shield=True):
 
                     # don't clobber an ongoing pdb
@@ -337,6 +340,8 @@ async def new_proc(
                         await do_hard_kill(proc)
 
                     log.debug(f"Joined {proc}")
+            else:
+                log.warning(f'Nursery cancelled before sub-proc started')
 
             if not cancelled_during_spawn:
                 # pop child entry to indicate we no longer managing this