fix(spammer): reject --mix weight totals that overflow u32

[kutluhaneth46]

Summary

Fixes #352.

--mix transfer=4294967295,legacy=1 parses successfully today because each weight is a valid u32, then panics in TxTypeMix::total_weight() during Config::validate() (attempt to add with overflow, exit 101).

Changes

• Add TxTypeMix::checked_total_weight() using checked_add.
• Reject overflowing aggregates in FromStr and Config::validate() with a clear --mix error.
• Make select_tx_type consume the checked total so generation cannot rely on overflowing arithmetic.
• Add regression tests for ordinary ratios, u32::MAX, multi-field overflow, and programmatic Config::validate() (error, not panic).

Test plan

• cargo test -p spammer tx_type_mix_ (and related config overflow test)

[huklaa]

@kutluhaneth46, this appears to reopen the same #352 implementation scope that you previously closed in #355 after acknowledging that my public claim came first.

For clarity, the chronology is already documented in #355: I claimed #352 first with the exact scope (overflow-safe aggregate handling, configuration-boundary rejection, and focused regression coverage), and you later wrote that my claim was visible first, that #355 covered the same scope, and that you were closing it so I could complete the work I had claimed.

PR #362 again says "Fixes #352" and implements substantially that same scope: checked aggregate addition, rejection in "FromStr" / "Config::validate", and overflow regression tests. Reopening the same implementation under a new PR number after explicitly closing the earlier duplicate does not resolve the coordination issue.

Please close #362 as well, or explain what materially changed in the chronology/scope since your acknowledgement on #355. I am continuing the work I publicly claimed first, and I’d like us to keep the contribution history and coordination transparent for maintainers.

[kutluhaneth46]

@huklaa You're right — thanks for catching this.

I had already closed #355 after acknowledging your earlier public claim on #352, and #362 should not have reopened that same implementation scope. Closing this PR now so you can finish the work you claimed first. Sorry for the noise.

[kutluhaneth46]

Closing as duplicate of the claimed #352 scope (see #355 / @huklaa).