Review forge lint "unsafe-typecast"

My brief examination found these were all false positives, but as a follow-up to https://github.com/FilOzone/filecoin-services/pull/555, we want to look over them to see if any are worth fixing. In some cases these truncations are intentional. In other cases they are harmless.

Do not add guards against impossible situations.

Assume that parameters supplied by the owner will be reasonable, though certain footguns may still be worth protecting against, depending on the impact.

Do not change function signatures.

Prefer to leave this lint disabled if there are more than 2 false-positives remaining in the codebase, rather than proliferating `forge-lint: disable-next-line(unsafe-typecast)`.

Start with the forge-lint unsafe-typecast findings related to follow-up PR #555 and inspect each reported cast for intentional truncation, harmless behavior, or a worthwhile fix. Done means addressing justified cases without guards for impossible situations or signature changes; if more than two false positives remain, leave the lint disabled rather than adding many suppression comments.

Written by the indexing model from the issue text.