Command-line debugger capabilities

[jeremyBanks]

We aren't have a fully-featured debugger, but we should add a few initial debugging features to our interpreter.

Currently, once per second we print the latest N (100?) instructions that we've executed, including their opcodes, addresses, clock time, and some debug info about what they modified. We throttle our execution so it doesn't exceed real-time, but aside from that we just run undisturbed until we hit a panic.

We should allow you to run in --debugger mode where you have commands like:

• next $N=1: Run next $N instructions.
• until $T: Run until clock time >= T.
• to $A: Run until instruction pointer equals $A.
• resume: Run forever.
• log: Displays the latest 20(?) instructions, and the next 8 bytes from the instruction pointer.
• registers: Show all registers.
• memory $N: Show value in memory at address $N.
• read $A: Displays the value in memory at address $A.

To make this most useful, we also need to:

• Update all methods that can panic so they instead return Results. If they return an Err, they must make sure to have had no side effects, so we can (if appropriate) bring up the debugger instead of aborting the program.

[jeremyBanks]

If we decided to implement some of these from the Web UI first, that's still good. It might be nice to step forward by frames...

• next vlbank

...while using the graphics debug panes.