atcab_write_enc corrupts block 0

[anpa8480]

Describe the bug
atcab_write_enc corrupts block 0 — When writing to an AES slot with WrCfg=4 (encrypted write), block 0 gets nonce-dependent garbage while block 1 writes correctly.

To Reproduce
Prerequisites: ATECC608B TrustFLEX, slot with WrCfg=4 (e.g. slot 5), IO protection key in slot 6.

Steps:

1. Write a known 16-byte key (e.g. all-zeros) to slot 5, block 0 via atcab_write_enc
   uint8_t key[16] = {0}; uint8_t io_key[32] = { /* IO protection key from slot 6 */ }; atcab_write_enc(5, 0, key, io_key, 6); // block 0

2. Read back block 0 via atcab_aes_encrypt_ext with known plaintext (NIST FIPS 197: PT=all-zeros, expected CT=66e94bd4ef8a2c3b884cfa59ca342b2e):
   uint8_t pt[16] = {0}, ct[16]; atcab_aes_encrypt_ext(atcab_get_device(), 5, 0, pt, ct); // ct does NOT match NIST expected — block 0 is corrupted

3. Repeat step 1 — block 0 produces different garbage each time (nonce-dependent).

4. Now write the same key to block 1:
   atcab_write_enc(5, 1, key, io_key, 6); // block 1

5. Verify block 1:

atcab_aes_encrypt_ext(atcab_get_device(), 5, 1, pt, ct);
// ct MATCHES 66e94bd4ef8a2c3b884cfa59ca342b2e — block 1 is correct

Expected behavior
Expected: Block 0 should produce NIST-matching output.
Actual: Block 0 contains nonce-dependent garbage; only block 1 writes correctly.

Additional context
ATECC608B TrustFLEX, RPi4, USB HID
cryptoauthlib version (v3.8.0)