Fix: missing cleanup on fail in print_report_xml_start

[mattmundell]

What

Use a simple fail jump instead of trying to do the cleanup at every return.

Why

This catches a few error exits that were not calling tz_revert and/or where not closing "out". Having so many separate returns was error prone.

Part of reducing the size of print_report_xml_start.

References

Follows /pull/2662.

Testing

I got a single report with the PR, then got the same report on the main branch. I made sure they were the same. I did that with two timezones just to be sure (using the timezone term in the filter).

o m m '<get_reports report_id="92a09ffd-a5bd-464b-9f2c-54b78ec5ebcf" details="1" filter="timezone=Pacific/Auckland rows=3 apply_overrides=1 min_qod=60"/>' > /tmp/r-auk-main
o m m '<get_reports report_id="92a09ffd-a5bd-464b-9f2c-54b78ec5ebcf" details="1" filter="timezone=Europe/Berlin rows=3 apply_overrides=1 min_qod=60"/>' > /tmp/r-ber-main
diff /tmp/r-auk /tmp/r-auk-main
diff /tmp/r-ber /tmp/r-ber-main