Always use blend for stem storage in cffwrite_t2cstr.c

[skef]

Description

The current version of this file stores stems in a union, with just a float for edge0 and edge1 in one case and a blend structure in the other. CFF2 uses the blend while CFF uses just the float. The blend .value happens to coincide with the float.

This sort of works but the model breaks down for non-blended stems in CFF2 output. It's better to just use the blend structure for all cases with hasBlend set to false if there's no blend data. That way CFF output can ignore everything but the "default" value and CFF2 output can correctly detect whether it needs to output a blend.

Checklist:

• I have followed the Contribution Guidelines
• I have added test code and data to prove that my code functions correctly
• I have verified that new and existing tests pass locally with my changes
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation

[cjchapman]

👍