icu4c: fix UTF-16 buffer under-allocation in vzone_write*

[Naveed8951]

Description

This PR fixes a heap under-allocation bug in the ICU4C VTimeZone C wrapper
write APIs (vzone_write, vzone_writeFromStart, vzone_writeSimple).

The wrapper computed resultLength using UnicodeString::length(), which
returns the number of UTF-16 code units, but allocated and copied only
resultLength bytes. Since char16_t is 2 bytes, the returned buffer was
smaller than implied by the returned length.

Because the API returns (char16_t*, length-in-code-units), callers may
legitimately read or write resultLength UTF-16 code units (including
adding a terminator), leading to heap out-of-bounds access.

Fix

• Allocate buffers in UTF-16 code units (sizeof(char16_t))
• Allocate (resultLength + 1) code units and explicitly NUL-terminate
• Copy resultLength UTF-16 code units
• Add overflow-safe size calculations
• Return U_MEMORY_ALLOCATION_ERROR on allocation failure

Impact

• Eliminates heap under-allocation and out-of-bounds access
• No change to API behavior or serialized output
• Improves robustness when handling untrusted VTIMEZONE input

Testing

• Added a regression test in intltest verifying:
  • u_strlen(out) == outLen
  • out[outLen] == 0

[CLAassistant]

All committers have signed the CLA.