[aysha-afrah26]

ExprAssignment::evaluate resolves the target entry with getEntry(key) before evaluating the right-hand side, and then locks it. When a string is assigned to an entry of another type, it looked the key up a second time through Blackboard::entryInfo(key) and called converter() on the result without checking it. If the key is unset between the two lookups (an UnsetBlackboard node or Blackboard::unset() on another thread sharing the blackboard, or an entry-update subscriber reacting to a write made by the right-hand side), entryInfo() returns nullptr and the process crashes with SIGSEGV. If the key is unset and created again in that window, the pointer refers to a different entry that only storage_ keeps alive, so that entry's converter is applied to the old one and a further unset can leave the pointer dangling. I ran into it with one thread executing x = '42' in a loop while another did unset("x") / set<int>("x", 1), and ASan reported the SEGV in operators.hpp on each of three runs. Reading entry->info from the entry that is already held and locked fixes it, which is also what the AnyTypeAllowed check a few lines above does. The change is header-inline with no API or ABI impact, and ParserTest.AssignStringWhileEntryIsUnset opens the window deterministically through a subscriber: it crashes on master and passes with the fix. The full ctest run is green apart from LoggerTest.SqliteLogger_ExecSqlStatement, which crashes the same way on master.