CData: address write_value issue in _parameters.csv

[rmcdermo]

This fixes the CData behavior reported in Discussion #2331 where Simple_parameters.csv was created with the expected headers and case names, but the randomized parameter values were blank.

Root cause: the CData parameter-output path had been disconnected in commit 7264729eb (CData: fixed error in write and removed unused code). In that change, the calls to write_value in create_case were commented out. Because those calls were no longer active, the write_value routine later appeared unused and was removed. The result was that CData still created the parameter CSV header row with add_header, but never copied each generated case’s sampled values into the row before flush_parameters_buffer.

The fix restores that missing path:

• Restored the type-bound write_value routine for value_wrapper_type.
• Re-enabled the write_value calls after each random fire and field is randomized in create_case.
• Kept the newer FIRE_DO_RAND behavior that avoids randomizing unassociated fire-generator points, so the previous CData crash fix remains intact.
• Changed CData’s initial input read to call read_input_file(.false.), so preprocessing does not create empty base-case CFAST output files such as Simple_vents.csv while running cdata Simple.in -P.

After this change, running:

cdata Simple.in -P