Enable use_alt_stack setting for crashtracker

I spent some time looking into `use_alt_stack` setting and I think
there's a bug in libdatadog and that setting doesn't seem to be doing
anything at this point.

Since the Ruby VM itself sets an altstack, this leaves us:

* In libdatadog 13.1, the crashtracker code always used the altstack
  (e.g. equivalent of `use_alt_stack` was hardcoded to `true`)
  => in libdatadog 13.1 we used the Ruby altstack
* In libdatadog 14.0, the crashtracker code chooses to use the altstack
  based on the value of `create_alt_stack` => libdatadog 14.0 we can't
  use the Ruby altstack

In preparation for `use_alt_stack` to get fixed in a future libdatadog
release, I'm changing it to true although, right now it's a no-op.

This means that in a future libdatadog release, we'll be back to where
we were in libdatadog 13.1: the crashtracker DOES NOT create its
own altstack, BUT WILL USE the existing Ruby altstack.

ext/libdatadog_api/crashtracker.c

@@ -67,7 +67,7 @@ static VALUE _native_start_or_update_on_fork(int argc, VALUE *argv, DDTRACE_UNUS
     // The Ruby crash handler also seems to get confused when this option is enabled and
     // "Process.kill('SEGV', Process.pid)" gets run.
     .create_alt_stack = false,
-    .use_alt_stack = false,
+    .use_alt_stack = true, // NOTE: This is a no-op in libdatadog 14.0; should be fixed in a future version
     .endpoint = endpoint,
     .resolve_frames = DDOG_CRASHT_STACKTRACE_COLLECTION_ENABLED_WITH_SYMBOLS_IN_RECEIVER,
     .timeout_ms = FIX2INT(upload_timeout_seconds) * 1000,