Make sure sorted and unsorted OTLP to STEF converters are consistent for PointValueTypeNone

[tigrannajaryan]

The sorted converter skips empty value type data points entirely (lines 121-122), while the unsorted converter processes them and maps them to PointValueTypeNone (baseotlptostef.go:46-47). This creates a behavior discrepancy between the two conversion paths where unsorted writes all data points but sorted skips empty ones. For consistency, consider either: 1) Having both paths skip empty value types, or 2) Having both paths convert them to PointValueTypeNone. The current implementation may cause confusion where the same input produces different output counts depending on which converter is used.

Originally posted by @Copilot in #371