fix(gnovm): correct type for shift expression

[ltzmaxwell]

shift operator where first operand is an untyped bigint always results in a bigint is not resolved by #1426, it's fixed by this one.

=================================================================

1. This is a fix to shift operator where first operand is an untyped bigint always results in a bigint #1462;
2. NOTE: This PR should be reviewed following the potential merger of feat: gno type check #1426, from which it is both decoupled and dependent. feat: gno type check #1426 serves as base branch of this one.
3. NOTE: Currently, this PR displays all code including that from feat: gno type check #1426, because it is being compared to the master branch instead of differing against feat: gno type check #1426 directly.

[deelawn]

Looks good to me. Thanks for quickly addressing my comments 🙏

[thehowl]

partial

[mvertes]

This PR greatly improves the behaviour of the shift operator. The tests are amazing and cover a lot of corner cases (several bugs also found in yaegi thanks to them) and match perfectly the results given by the compiler. Thank you for this effort. My remarks are minor and do no block approval. We should merge it ASAP now.

[mvertes]

I see that some tests are not passing, even if they print the correct results. This is due to eval_tests not trimming spaces in results. Let me fix this.

[thehowl]

thanks for the speedy merge @mvertes! <3