LessThanOrEquals needs a sanity check on bitlength?

[rubensayshi]

It seems like there's some magical threshold for the bitlength of the numeric context and the bitlength of the modulus at which LessThanOrEquals breaks.

I thought that the bitlength of the modulus would have to be >= 2x bitlength of the numeric context, but it doesn't seem that way, at higher sizes the bitlength can be much bigger than 1/2 the modulus and at lower sizes the bitlength needs to be much smaller than 1/2 the modulus...

There's probably a logical reason / relation and it might be good to have a sanity check there.

these 2 tests both fail (while 453 / 512 for example works fine);

  @Test
  public void test_compareLt_Sequential454_512() {
    TestParameters tp = new TestParameters()
      .field(new MersennePrimeFieldDefinition(512, 569))
      .maxBitLength(454);

    runTest(new CompareTests.TestCompareLT<>(), tp);
  }

  @Test
  public void test_compareLt_Sequential32_64() {
    TestParameters tp = new TestParameters()
      .field(new MersennePrimeFieldDefinition(64, 569))
      .maxBitLength(32);

    runTest(new CompareTests.TestCompareLT<>(), tp);
  }

[rubensayshi]

another odd thing;

if you do compareLEQ(0, 2 ^ maxBitLength) then the result is 2