Fix empty hlsfactory.toml for DGNN_Booster_convLSTM + add config audit script

[CharithKumarGunda]

while exploring issue #69 (New Design Configuration), I wrote the script (audit_configs.py) to check which designs have a valid hlsfactory.toml file and validate them using read_design_config.

This audit found that hlsfactory/hls_dataset_sources/accelerators/DGNN_Booster_convLSTM/hlsfactory.toml was an empty (0-byte) file, causing a DesignConfigError: design_name is required.

I fixed this by acutally creating a proper hlsfactory.toml based on the existing dataset_hls.tcl entry point and the format used by other designs (e.g. soda__blur).

The audit script also shows which datasets still need hlsfactory.toml files created (polybench, machsuite, chstone, forgebench, and others) - happy to help with this migration if useful!

[CharithKumarGunda]

Update: I mainaged to also migrated the full PolyBench dataset to hlsfactory.toml config files (all 9 designs: atax, bicg, gemm, gesummv, k2mm, k3mm, mvt, syr2k, syrk). Each design uses the same flow pattern (VitisHLSSynthFlow, VitisHLSCosimSetupFlow, VitisHLSImplFlow) based on their existing dataset_hls.tcl files. I made sure to also validated all 9 with read_design_config() - and found no errors.

Ran audit_configs.py again to confirm: polybench now shows 9/9 designs with valid configs.

This PR now includes both the DGNN_Booster_convLSTM fix and the full PolyBench migration. Happy to continue with MachSuite and CHStone next if this approach looks good!