Distinguish memory read failures from corrupt method tables in SOS error messages (#5680)

SOS commands like `!dumpheap -stat` throw identical error messages for
two distinct failure modes: memory read failures vs. successfully
reading a zero/invalid MethodTable. These require different diagnostic
approaches.

NOTE: LLDB reports missing memory as if it were zeroed memory so SOS is still going
to claim corrupt method tables in place of missing memory there.

Co-authored-by: noahfalk <[email protected]>

Author: Copilot

src/SOS/Strike/sos.cpp

@@ -103,10 +103,10 @@ namespace sos
         {
             TADDR temp;
             if (FAILED(MOVE(temp, mAddress)))
-                sos::Throw<DataRead>("Object %s has an invalid method table.", DMLListNearObj(mAddress));
+                sos::Throw<DataRead>("Failed to read memory at Object %s.", DMLListNearObj(mAddress));
 
             if (temp == (TADDR)0)
-                sos::Throw<HeapCorruption>("Object %s has an invalid method table.", DMLListNearObj(mAddress));
+                sos::Throw<HeapCorruption>("Object %s has a corrupt method table.", DMLListNearObj(mAddress));
 
             mMT = temp & ~METHODTABLE_PTR_LOW_BITMASK;
         }