chore: Only resolved globals monomorphization

[vezenovm]

Description

Problem*

Small change noticed while working on globals

Summary*

We should only allow accessing resolved globals in monomorphization after switching to computing all globals at compile time.

Additional Context

Documentation*

Check one:

• No documentation needed.
• Documentation included in this PR.
• [For Experimental Features] Documentation to be submitted in a separate PR.

PR Checklist*

• I have tested the changes locally.
• I have formatted the changes with Prettier and/or cargo fmt on default settings.

[github-actions]

Compilation Report

Program	Compilation Time	%
sha256_regression	1.294s	-1%
regression_4709	0.792s	-2%
ram_blowup_regression	18.140s	-1%
rollup-root	3.880s	-10%
rollup-block-merge	4.660s	11%
rollup-base-public	35.800s	1%
rollup-base-private	11.600s	-11%
private-kernel-tail	1.002s	2%
private-kernel-reset	8.394s	0%
private-kernel-inner	2.322s	4%

[github-actions]

Execution Report

Program	Execution Time	%
sha256_regression	0.054s	0%
regression_4709	0.001s	0%
ram_blowup_regression	0.575s	0%
rollup-root	0.106s	1%
rollup-block-merge	0.104s	-1%
rollup-base-public	1.220s	0%
rollup-base-private	0.463s	0%
private-kernel-tail	0.021s	0%
private-kernel-reset	0.348s	0%
private-kernel-inner	0.078s	1%

[github-actions]

Compilation Memory Report

Program	Peak Memory
keccak256	78.11M
workspace	123.80M
regression_4709	422.96M
ram_blowup_regression	1.58G
rollup-base-public	4.11G
rollup-base-private	1.20G
private-kernel-tail	207.17M
private-kernel-reset	730.59M
private-kernel-inner	294.40M

[github-actions]

Execution Memory Report

Program	Peak Memory
keccak256	74.65M
workspace	123.81M
regression_4709	315.98M
ram_blowup_regression	512.41M
rollup-base-public	480.66M
rollup-base-private	326.51M
private-kernel-tail	181.02M
private-kernel-reset	249.69M
private-kernel-inner	209.86M

[jfecher]

I don't remember if this was originally included as part of the change to lazily elaborate globals and whether there were cases with comptime code where we could run into unresolved globals during monomorphization like this or not. Since it passes our tests though, and I agree we ideally should have all globals be resolved by here, it LGTM.