redo-log: prioritize the "foreground" process.

When running a parallel build, redo-log -f (which is auto-started by
redo) tries to traverse through the logs depth first, in the order
parent processes started subprocesses.  This works pretty well, but if
its dependencies are locked, a process might have to give up its
jobserver token while other stuff builds its dependencies.  After the
dependency finishes, the parent might not be able to get a token for
quite some time, and the logs will appear to stop.

To prevent this from happening, we can instantiate up to one "cheater"
token, only in the foreground process (the one locked by redo-log -f),
which will allow it to continue running, albeit a bit slowly (since it
only has one token out of possibly many).  When the process finishes,
we then destroy the fake token.  It gets a little complicated; see
explanation at the top of jwack.py.

state.py

@@ -21,6 +21,8 @@ ALWAYS='//ALWAYS'   # an invalid filename that is always marked as dirty
 STAMP_DIR='dir'     # the stamp of a directory; mtime is unhelpful
 STAMP_MISSING='0'   # the stamp of a nonexistent file
 
+LOG_LOCK_MAGIC=0x10000000  # fid offset for "log locks"
+
 
 class CyclicDependencyError(Exception): pass
 
@@ -374,9 +376,11 @@ class Lock:
             self.owned = True
         return self.owned
 
-    def waitlock(self):
+    def waitlock(self, shared=False):
         self.check()
-        fcntl.lockf(self.lockfile, fcntl.LOCK_EX, 0, 0)
+        fcntl.lockf(self.lockfile,
+            fcntl.LOCK_SH if shared else fcntl.LOCK_EX,
+            0, 0)
         self.owned = True
             
     def unlock(self):