CORE-5529 Incorrect Results from Evaluation of Filter Expressions with Float Members is not resolved.

Per the 5.1.0 release notes:

3.2.2 Incorrect Results from Evaluation of Filter Expressions with Float Members

Evaluation of filter expressions that referred to float members may have had erroneous results. For example:

struct MyType {
   float m1;
};

Evaluating the SQL filter expression "m1 = 4.1" for the above type may have caused an erroneous result. Therefore a sample that should have passed a filter may have been filtered out.

However the fix referenced in the release notes does not appear to correct the issue in the case of nested structures.

struct MyNestedType {
    float m1;
};
 
struct MyType {
    MyNestedType m_struct; 
};

The above IDL definitions cause filter failures with similiar resuts on the expression "m_struct.m1 = 4.1".  We have also seen problems with filtering on longs inside nested structures as well, but in this case, samples which should not pass the filter end up being permitted.

 --Jason C. Wenger