guest memory: place stack below text/code

[SchmErik]

The previous memory layout was prone to a security vulnerability because the
stack could collide with the heap. This change places the stack below
TEXT_START with enough space for 2 MB of stack. This also results in increased
locality of memory which will result in less page faults for small programs.

[SchmErik]

@intoverflow I'm not sure if 2MB is large enough for zeth... Feel free to chime in and let us know if it's not enough!

[github-actions]

Benchmark for Linux-cuda 29d0914

Click to hide benchmark

Test	Base	PR	%
fib/100/execute	5.2±0.11ms	5.2±0.08ms	0.00%
fib/100/prove	791.8±1.66ms	724.2±5.75ms	-8.54%
fib/100/total	796.2±2.72ms	726.0±3.98ms	-8.82%
fib/1000/execute	5.9±0.11ms	5.8±0.13ms	-1.69%
fib/1000/prove	818.7±3.01ms	755.4±4.46ms	-7.73%
fib/1000/total	824.5±2.27ms	759.6±5.03ms	-7.87%
fib/10000/execute	12.1±0.11ms	11.7±0.15ms	-3.31%
fib/10000/prove	2.8±0.01s	2.5±0.01s	-10.71%
fib/10000/total	2.8±0.01s	2.5±0.01s	-10.71%

Benchmark for Linux-default 29d0914

Click to hide benchmark

Test	Base	PR	%
fib/100/execute	6.7±0.46ms	6.6±0.44ms	-1.49%
fib/100/prove	5.1±0.03s	5.0±0.02s	-1.96%
fib/100/total	5.1±0.03s	5.0±0.02s	-1.96%
fib/1000/execute	7.8±0.51ms	7.6±0.48ms	-2.56%
fib/1000/prove	5.1±0.03s	5.1±0.02s	0.00%
fib/1000/total	5.1±0.03s	5.1±0.03s	0.00%
fib/10000/execute	14.3±0.48ms	14.0±0.47ms	-2.10%
fib/10000/prove	20.9±0.10s	20.9±0.08s	0.00%
fib/10000/total	21.0±0.10s	20.9±0.15s	-0.48%

Benchmark for macOS-default

    <details open>
      <summary>Click to hide benchmark</summary>
      Benchmarks have changed between the two branches, unable to diff.
    </details>

Benchmark for macOS-metal

    <details open>
      <summary>Click to hide benchmark</summary>
      Benchmarks have changed between the two branches, unable to diff.
    </details>

[Wollac]

I'm not sure if 2MB is large enough for zeth... Feel free to chime in and let us know if it's not enough!

There is a test https://github.com/risc0/zeth/blob/main/testing/ef-tests/tests/executor.rs that attempts to run an artificial transaction that should require a lot of stack in the guest. It might be worth checking these changes against this test. If it passes, I think we are pretty sure that it should work for real Ethereum blocks as well.