Use M.Q.Convert.DoubleAsStringWithPrecision to trim Doubles printed in the katas

Follow-up to #1406: now that the function M.Q.Convert.DoubleAsStringWithPrecision is available, we can use it in the katas to shorten the long floating-point numbers.

Here are a few examples of the katas that print long doubles:

* Complex Arithmetic
* Single-Qubit Gates kata, task "Prepare Arbitrary State"
* Preparing Quantum States, task "Unequal Superposition"

There are probably others; the typical scenario in which this occurs is printing the Double parameters of the test or returns of the solution for an incorrect solution. We are unlikely to need more than 2-3 digits of precision in these cases.

Read M.Q.Convert.DoubleAsStringWithPrecision, then inspect the Complex Arithmetic, Single-Qubit Gates, and Preparing Quantum States katas for output that includes Double values. Run the affected kata checks and confirm incorrect-solution messages display shortened values with roughly 2–3 digits of precision.

Written by the indexing model from the issue text.