ibc: Chain ID parsing incorrectly accepts newlines

[avahowell]

Describe the bug
When parsing chain IDs, ibc-types uses a more lenient regex than ibc-go's regex.

For example, ibc-go will consider the chain ID \n-foo-1 to have revision number 0, whereas penumbra-ibc will consider it to have revision number 1.

This likely does not have any security impact, but it is a deviation from the specification and can potentially lead to an issue in the future.

Expected behavior
Include a newline (and possibly anchors) in ibc-types's version of the regex.

pub fn is_epoch_format(chain_id: &str) -> bool {
  - let re = safe_regex::regex!(br".*[^-]-[1-9][0-9]*");
  + let re = safe_regex::regex!(br"^.*[^\n-]-[1-9][0-9]*$");
       re.is_match(chain_id.as_bytes())
   }

[conorsch]

We didn't fix this as an immediate result of the audit. It'd be nice to clean it up, but we should inspect where the chain id is used, particularly in IBC contexts like handshakes. Marking this issue as consensus-breaking so that a fix doesn't get shipped without upgrade/migration plans.