…nce(): Use Assert.AreEqual instead of Asssert.IsTrue with == to fix floating point comparison on x86 .NET Framework with optimizations enabled.

…NumericUtils.SingleToSortableInt32() to prevent test failures on x86 .NET Framework. This fixes Lucene.Net.Search.TestBooleanQuery.TestBS2DisjunctionNextVsAdvance(), Lucene.Net.Search.TestFuzzyQuery.TestTieBreaker(), and Lucene.Net.Sandbox.Queries.TestSlowFuzzyQuery.TestTieBreaker(). See apache#269.

….Util.NumericUtils.SingleToSortableInt32() to prevent test failures on x86 .NET Framework. This fixes Lucene.Net.Search.TestTopDocsMerge::TestSort_1(), Lucene.Net.Search.TestTopDocsMerge::TestSort_2(), and Lucene.Net.Search.TestSearchAfter.TestQueries(). See apache#269.

… using Lucene.Net.Util.NumericUtils.SingleToSortableInt32() to prevent test failures on x86 .NET Framework. This fixes Lucene.Net.Sandbox.Queries.TestSlowFuzzyQuery.TestTokenLengthOpt(). See apache#269.

…tils.SingleToSortableInt32() to prevent test failures on x86 .NET Framework. See apache#269.

…e floating point numbers (except in cases where we are testing for whole numbers).

…ng float to double loses precision in x86 .NET Framework. Do an intermediate cast to decimal to fix Lucene.Net.Expressions.TestExpressionSorts::TestQueries(). See apache#269.

…c.Comparer<double> to ensure we use the same comparison logic as in Java.

…est jobs