Fix panic in gen_indexes when index modulo N equals zero

[mwaddip]

Summary

gen_indexes panics when any of the 32 extracted 4-byte hash windows,
interpreted as an unsigned big-endian integer, is an exact multiple of N.
The modulo result is zero, BigInt::to_u32_digits() returns an empty
digit vector, and .1[0] panics with index out of bounds.

The JVM reference (AutolykosPowScheme.scala) handles this correctly
via BigInt.toInt, which returns 0 for BigInt(0).

Fix

Replace .1[0] with .1.first().copied().unwrap_or(0) — returns 0
when the digit vector is empty, matching JVM behavior.

Risk

With 32 indices per header and N starting at ~67M, there is a non-trivial
cumulative probability that some mainnet block triggers this during
initial chain sync. The fix is a one-line change with no behavioral
difference for non-zero results.

Test plan

• Added regression test test_gen_indexes_zero_modulo that triggers
  the exact panic condition (seed hash with 4-byte window equal to N)
• All existing tests pass unchanged
• cargo clippy clean
• cargo fmt clean

[coveralls]

Pull Request Test Coverage Report for Build 23799568905

Details

• 17 of 17 (100.0%) changed or added relevant lines in 1 file are covered.
• 4 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.01%) to 86.904%

---

Files with Coverage Reduction	New Missed Lines	%
ergo-lib/src/wallet/tx_builder.rs	4	92.98%

Totals
Change from base Build 21705206395:	0.01%
Covered Lines:	27486
Relevant Lines:	31628

---

💛 - Coveralls