Revisit in-failure and in-stall console dumps of scheduler state, plus few other misc. things

• remove doubling console dumps of the scheduler state occurring in a failure/stall;
• [.. and minor cleanup of associated scheduler's state-dumping logging messages;]
• improve the coverage of logging protection by msgLevel(X);

And several other things:

• re-implement PRG::accept() to walk visitors through all nodes;
• make StatusCode state checking explicit;
• remove unused and obsolete headers.