Reevaluate IL emitting for symbolic data analysis solutions

[gkronber]

Could be beneficial for larger datasets. Additionally it could be beneficial for solutions that are evaluated multiple times.

[gkronber]

Resurrected the IL emitting interpreter I once implemented for HL 3.1 and implemented an interpreter for 3.4 based on it. The unit test tests the same cases as the unit test for the standard interpreter except for the ADF operations which are not implemented in the IL emitting interpreter.

[gkronber]

r6732:

• added IL emitting tree interpreter for symbolic data analysis and test case.
• Found and fixed a bug in the existing interpreter for boolean operators OR and AND with NaN arguments. However, this means that the output of previously stored solutions changes. We Probably we should keep the incorrect behavior now and document this accordingly, as changing this would need a version increment of all data analysis plugins.

[gkronber]

r6735: reduced nodesPerSecThreshold for IL-emitting interpreter performance tests to fix problems with the release build on builder.

[gkronber]

r6755: implemented code to handle root symbols for the il emitting interpreter and fixed code for power symbol.

[gkronber]

r6770: added code for the interpretation of lagged variables to the IL emitting interpreter.

[gkronber]

r6809:

• Added check if row index lies within the possible range of the dataset and implemented interpretation of time series symbols in the IL emitting interpreter.
• Added test cases for evaluation of time series symbols.

[gkronber]

r6836: reduced threshold for evaluated tree nodes / sec for the IL emitting interpreter performance test to reduce number of fails on the builder.

[gkronber]

r6849: experimental change to potentially improve performance of interpreters.

[gkronber]

r6860: reverse merged changes to SymbolicDataAnalysisExpressionTreeInterpreter.cs in r6849 as there were no performance improvements observable for this interpreter.

[mkommend]

I've tested the whole functionality and had a brief look at the source code. Everything works as expected, but one thing should be corrected:

• Throw a more meaningful exception, if symbols are not implemented in the ILEmittingInterpreter (e.g., specifying which symbol causes the exception).

[gkronber]

r7077: added a message for the NotSupportedException thrown in the SymbolicDataAnalysisExpressionTreeILEmittingInterpreter stating which symbolic is not supported.

[mkommend]

Thx for updating the text of the thrown exceptions.