Time of lalpulsar_MakeSFTs is significantly slower now

This could be related to recent changes in lalpulsar_MakeSFTs related to frame reading, but I am finding that the time to create SFTs using the new v3 specification in lalpulsar-6.0 is significantly slower than before. Is this an expected outcome of the recent updates to lalpulsar?

For example, using lalpulsar=5.2.0:

$ time lalpulsar_MakeSFTs -f 7 -t 30 -p . -C frm.cache -s 1364056500 -e 1364056530 -N H1:PEM-EX_MAG_VEA_FLOOR_X_DQ -F 10 -B 190 -w 3 -P 0.5 -u ADC_REAL4
lalpulsar_MakeSFTs -f 7 -t 30 -p . -C frm.cache -s 1364056500 -e 1364056530 -  0.49s user 0.09s system 84% cpu 0.684 total

And using lalpulsar=6.0.0.1:

time /opt/lscsoft-kw/bin/lalpulsar_MakeSFTs -O 100 -K AUX -R 1 -f 7 -t 30 -p . -C frm.cache -s 1364056500 -e 1364056530 -N H1:PEM-EX_MAG_VEA_FLOOR_X_DQ -F 10 -B 190 -w hann -P 0.5
2023-03-30 10:43:01.1445 (39349) [normal]: Parsed user input successfully
2023-03-30 10:43:50.7141 (39349) [normal]: Generated first SFT at GPS time 1364056500
2023-03-30 10:43:50.7142 (39349) [normal]: Generated last SFT at GPS time 1364056500
2023-03-30 10:43:50.7142 (39349) [normal]: Generated 1 SFTs
/opt/lscsoft-kw/bin/lalpulsar_MakeSFTs -O 100 -K AUX -R 1 -f 7 -t 30 -p . -C   40.06s user 9.28s system 99% cpu 49.662 total

Essentially, it took ~0.5 s to create a v2 SFT, whereas it took ~40s to create a v3 SFT. This was tested locally on my MacBook Pro, using a compiled version of lalpulsar from @karl-wette's branch in !2118 (merged).