fix: guard parsing of hex words against missing 0x prefix

[strmfos]

Describe your changes

• prevent panic in InputFile::parse_word when inputs omit the 0x prefix
• return descriptive error for malformed word strings and keep behavior otherwise unchanged
• verified with cargo test internal::tests::test_merkle_data_parsing and cargo test miden_vm::internal::tests::test_merkle_data_parsing

[huitseeker]

The fix to use strip_prefix() is correct, but this change needs proper test coverage.

It would be great if you could add unit tests to the test_merkle_data_parsing module (around line 370) to verify:

1. Error handling for missing 0x prefix
2. Edge cases like empty strings or just 0x
3. Valid hex parsing still works as expected

[strmfos]

Thanks for the review! Added the requested unit tests in test_merkle_data_parsing module:

• test_parse_word_missing_0x_prefix() - verifies error handling for missing "0x" prefix
• test_parse_word_edge_cases() - covers empty strings, just "0x", and short hex
• test_parse_word_valid_hex() - ensures valid hex parsing still works

[huitseeker]

Thanks!

[bobbinth]

Theoretically, we should be able to replace this function with Word::try_from() - though, this would not support "incomplete strings" (i.e., the string would need to be 66 characters). Maybe it is worth updating how Word::try_from() works in miden-crypto? cc @huitseeker.