InvalidChecksum fields disagree with their documentation

{
  "title": "InvalidChecksum fields disagree with their documentation",
  "body": "## Description\n\n`InvalidChecksum` returns field values in the reverse order from their documentation.\n\n## Steps to Reproduce\n\nWith bs58 0.5.1 and the `check` feature enabled, run this example. Its success reproduces the discrepancy:\n\n```rust\nlet input = bs58::encode([42, 0, 0, 0, 0]).into_string();\nassert_eq!(\n    bs58::decode(input).with_check(None).into_vec(),\n    Err(bs58::decode::Error::InvalidChecksum {\n        checksum: [255, 18, 44, 14],\n        expected_checksum: [0, 0, 0, 0],\n    })\n);\n```\n\n## Expected Behavior\n\nThe field documentation describes `checksum` as supplied and `expected_checksum` as calculated.\n\n## Actual Behavior\n\nThe supplied checksum is `[0, 0, 0, 0]`; the calculated checksum is `[255, 18, 44, 14]`. The returned values are reversed. The same mismatch occurs with `cb58`. Invalid checksums are still rejected, and the `Display` implementation follows the returned field order; code interpreting the structured fields according to their documentation receives reversed diagnostics.\n\n## Environment\n\nbs58 0.5.1 with `check`; the analogous path uses `cb58`.",
  "validity": "reported",
  "upstream_status": "not_reported"
}