Fix for potential typos

[lanceXwq]

Hi! I noticed a potential documentation issue in the od help/manpage:

• -f is described as “double precision (32-bit)”, aren't 32-bit floats single precision?
• -i is described as “decimal 8-byte units”; if I’m not mistaken, -i corresponds to C int (typically 4-byte) / -t dI.

This PR updates the wording accordingly. No behavior changes—docs/help only. Thanks!

[github-actions]

GNU testsuite comparison:

Note: The gnu test tests/tail/tail-n0f is now being skipped but was previously passing.

[ChrisDryden]

Let me know if you think my comment about the -i vs -I is correct but the -f definitely seems correct to me just might be an issue with the french translation

[ChrisDryden]

Thanks!