Update Padding from 0x1 to 0x6 as in SHA3 standard

[zeruniverse]

I observed CryptoJS SHA3 is different from PHP hash("sha3-512"). After comparing to code here: https://github.com/emn178/js-sha3 I found changing the 0x1 to 0x6 will make CryptoJS matching the standard SHA3 (at least for sha3-512)

[alanturing881]

Independently reproduced this while auditing crypto-js — confirming and extending this fix with broader test coverage.

Verification

Differential-tested CryptoJS.SHA3() against Node.js's native crypto.createHash('sha3-*') across all four configured output sizes (224/256/384/512), at 31 lengths spanning each size's Keccak rate boundary (72/104/136/144 bytes respectively), including the empty string:

• Before this fix (current 0x1 padding byte): 0/31 matches for every output size (SHA3-224, SHA3-256, SHA3-384, SHA3-512) — CryptoJS.SHA3('') never matches the standard SHA-3 digest of the empty string, or any other input.
• With this PR's one-line fix (0x1 → 0x6): 14/14 matches exactly against Node's native sha3-512 in a follow-up sweep.

This confirms the domain-separation padding byte is the sole root cause, and that the fix is correct not just for SHA3-512 (as originally reported) but for all four SHA-3 output sizes this library exposes via outputLength.

Impact note

To be clear about severity: this isn't a cryptanalytic weakness (Keccak with 0x01 padding isn't known to be weaker than FIPS 202 SHA-3 with 0x06 padding — both share the same underlying permutation and security margins). The practical impact is interoperability: CryptoJS.SHA3() currently computes the pre-standardization Keccak submission, not NIST-standardized SHA-3, so it silently produces digests that don't match any other SHA-3 implementation (OpenSSL, Node's native crypto, Python's hashlib, etc.) or FIPS-140/202 compliance requirements — for every single input.

This PR has been open and mergeable: true / mergeable_state: clean since 2020. Given it's a single-character, fully backward-compatible-with-the-standard fix that resolves a real conformance defect affecting every CryptoJS.SHA3() call, it'd be great to see this merged.