Add VVC-VTM-AS preset and bug-fixes to csv-export

[vibhoothi]

This patchset enables AS preset for VVC along with various bug fixes to CSV_Export,

• Do not lower-case when writing This creates a problem for Class E where two sequences were flipped
• Explicit cast of perf-stat when writing to CSV/XLSM file This avoids treating the last col as Text
• Do not write A1-4K for LD
• Write VVC-VTM AS config

Depends on https://github.com/xiph/rd_tool/pull/145

[vibhoothi]

The AS support for VVC is still in the Preliminary stage, because of a few reasons

• We have to modify the convex-hull framework [preferably upstream in AVM] for handling non-AVM code paths there to reduce divergence from AVM toolkits, as VVC QPs are entirely different.
  • Due to this reason, we only have the AS generation in the XLSM template and not in the Frontend as more interest of CWG?
  • Frontend will be working as such only difference from AVM AS will be WebUI-based convex-hull rendering which will be missing for VVC-based runs.
• Tested with one video in AS preset in VLC12 because of storage reasons, VVC requires input to be YUV (so source to be converted to YUV per QPs) and later whole pipeline is in YUV. So this is a problem for running full AS [or even full AOM_CTC even for VVC at one stretch] since we have only ~120GB of free space per-runners.

We do get points and so, in theory, it should work, for full-fledged testing current resources are not sufficient.

Note: The numbers look off as comparing 1F AVM with 130F VVC, only showing this for demonstration purposes.

Edit: If we have 500GB extra per-each runner we should be fine for full CTC for VVC given that we pre-render YUV files of source and store it in same path as other y4m files (~250GB) and an extra ~250gb for YUV-based pipeline handling(here alone we might need around +500gb per-machine)