Update debuginfo dexter stepping for new lldb behavior

dexter sets breakpoints on every source line in the file.
Then it step-in's to every source line, and tests are written
to expect that one step will move between lines, hitting the
breakpoints that are set there.

With this new algorithm, lldb will now step to the next source
line, and be stopped at a breakpoint, but it has not yet hit the
breakpoint.  The stop reason will be "step-in completed".  When
you do "step-in" again, we will hit the breakpoint instruction,
stop with a "breakpoint-hit" stop reason -- but the pc has not
advanced.  Then when you "step-in" again, we will move to the
next source line.

In short, lldb needs to step twice in a situation with breakpoints
on every source line and stepping across them.

This patch detects when lldb is behaving this way, and preserves
the behavior the tests expect.  It will step-in, and if we stop
with a step-in stop reason and we are stopped at a breakpoint site,
we will step once again to hit the breakpoint.

Author: jasonmolenda

cross-project-tests/debuginfo-tests/dexter/dex/debugger/lldb/LLDB.py

@@ -206,6 +206,28 @@ def launch(self, cmdline):
 
     def step(self):
         self._thread.StepInto()
+        stop_reason = self._thread.GetStopReason()
+        # If lldb has stepped to a breakpoint, but not
+        # executed it yet, lldb used to report eStopReasonBreakpoint
+        # even though the breakpoint hadn't been hit.
+        # Now lldb will report eStopReasonPlanComplete for
+        # the step completing, and when you step again, you will
+        # hit the breakpoint at this location -- because you have
+        # now actually executed the breakpoint instruction.  This
+        # means that stepping past a source line with a breakpoint
+        # on the first instruction requires stepping twice.  All of
+        # the debuginfo tests assume that it will succed with a single
+        # step.  So we silently step again to hit the breakpoint, when
+        # we've stopped at a breakpoint location but haven't hit it yet.
+        if stop_reason == self._interface.eStopReasonPlanComplete:
+            stepped_to_breakpoint = False
+            pc = self._thread.GetFrameAtIndex(0).GetPC()
+            for bp in self._target.breakpoints:
+                for bploc in bp.locations:
+                    if bploc.GetAddress().GetLoadAddress(self._target) == pc:
+                        stepped_to_breakpoint = True
+            if stepped_to_breakpoint:
+                self._thread.StepInto()
 
     def go(self) -> ReturnCode:
         self._process.Continue()