Fix segfault when accessing fields/field_types on freed result

[jonathan-biskycreative]

Summary

Fixes segfault when accessing .fields or .field_types on a Result object after the internal result has been freed. This is particularly common with 0-row query results.

Problem

A segfault occurs in the following sequence:

1. Query returns 0 rows
2. Internal row caching iterates (0 iterations), never populating wrapper->fields
3. Result is freed after iteration completes (0 processed == 0 total)
4. .fields is called, wrapper->fields is Qnil
5. mysql_num_fields(wrapper->result) accesses freed memory
6. SEGFAULT

Fix

This PR includes two improvements:

1. Eager field population (primary fix)

Fields are now populated immediately when a result is created, before any iteration can free the result:

// In rb_mysql_result_to_obj
if (r != NULL && !wrapper->is_streaming) {
  rb_mysql_result_fetch_fields(obj);
}

This prevents the problematic scenario from occurring at all.

2. Safety check (fallback)

The resultFreed check (which already exists in other functions like rb_mysql_result_each_) is now added to the fields accessor functions:

if (wrapper->resultFreed) {
  rb_raise(cMysql2Error, "Result set has already been freed");
}

If an edge case still occurs, this raises a catchable Ruby exception instead of crashing the process.

Testing

This fix has been tested in a production Rails 8.0 application where the segfault was occurring ~9 times per month during queries that return 0 rows. With the eager field population, the issue no longer occurs.

Fixes #1426