Add vstrtonum util as replacement for atof/strtod, etc.

[markcmiller86]

Description

This is a draft PR for developers to see where I am headed with this. It defines vstrtonum<sometype>() with some special sauce to handle things like default values, error checking, range checking, etc. Read the comment in StringHelpers.h for an overview...

visit/src/common/utility/StringHelpers.h

Lines 93 to 160 in dc5eefb

	// ****************************************************************************
	// Function: vstrtonum
	//
	// Purpose: Replacement for strtoX() and atoX() methods.
	//
	// Instead of any of these kinds of uses...
	//
	// int k = atoi(numstr);
	// float f1 = atof(numstr);
	//
	// unsigned u = (unsigned) strtoul(numstr, 0);
	// if (errno != 0) u = 0xFFFFFFFF; // set to default val
	//
	// float f2 = (float) strtod(numstr, 0);
	// if (errno != 0) {
	// f2 = -1.0; // set to default value
	// debug5 << numstr << " bad value" << endl; // log error
	// }
	//
	// ...do this...
	//
	// int k = vstrtonum<int>(numstr);
	// float f1 = vstrtonum<float>(numstr);
	//
	// unsigned u = vstrtonum<unsigned>(numstr, 0xFFFFFFFF);
	// float f = vstrtonum<float>(numstr, -1.0, debug5);
	//
	// Templatized methods to convert strings to language-native typed
	// numeric values, perform some minimal error checking and optionally
	// emit error messages with potentially useful context when errors
	// are encountered.
	//
	// This should always be used in place of strtoX() or atoX() when
	// reading ascii numerical data.
	//
	// We do a minimal amount of error checking for a signed conversion
	// by checking if first non-whitespace character is a minus sign and
	// artificially setting errno to EDOM (not something strtoX/atoX
	// would ever do). We could add more error checking for different
	// cases too by, for example, checking value read for int type and
	// seeing if it is too big to fit in an int. We currently do not
	// do this but it would be easy to add. We could also easily add
	// logic for other, less frequently used types such as shorts or
	// maybe int64_t, etc.
	//
	// The default method treats all ascii as long double for the strtoX
	// conversion and then casts it to correct type. We specialize some
	// cases for slightly better behavior.
	//
	// I ran performance tests on macOS doing 1 million conversions with
	// these methods (including error checking) and 1 million directly
	// with strtoX and atoX methods and observed no significant diffs
	// in performance. In addition, keep in mind that these methods are
	// typically being used in conjunction with file I/O, which almost
	// certainly dominates performance. The only time this might not be
	// true is for memory resident "files", mmaps, and/or SSDs.
	//
	// At some point, it would make sense to enhance this to use locale
	// so we can handle different character encodings as well as regional
	// specific interpretations (e.g. European 1.234.456,89). The idea
	// would be to set the locale VisIt is using (a pref. maybe) and then
	// these methods would just use that locale in calls to strtoX. That
	// could be achieved globally in VisIt with a call to setlocale().
	// However, when in the United States reading an ascii data file
	// formatted for human readability in Germany the desire would be
	// to specify "de_DE" for the locale during the read of just that
	// file suggesting something more complicated than just a global
	// setting.