Add advanced instruct questions

[wongjingping]

Add advanced instruct questions.
Our columns include "instructions" (which is meant for inserting into the prompt, as well "correct_instructions" (which is the actual instruction that the model is supposed to follow, added only for easier reference/debugging).

Some sample benchmarks:

python3 main.py \
  -db postgres \
  -q data/instruct_advanced_postgres.csv \
  -o results/openai_gpt4_turbo_advanced.csv \
  -g oa \
  -f prompts/prompt_openai.md \
  -m gpt-4-0125-preview \
  -c 0 \
  -p 20
Preparing questions...
Using all questions from data/instruct_advanced_postgres.csv
Correct so far: 37/64 (57.81%): 100%|███████████████████████████████████████████████████████████████████████████████████████████| 64/64 [00:44<00:00,  1.45it/s]
                 query_category  num_rows  mean_correct  mean_error_db_exec
0         instructions_cte_join        16         0.625              0.0625
1       instructions_cte_window         8         0.375              0.2500
2        instructions_date_join        16         0.500              0.1250
3  instructions_string_matching         8         0.875              0.0000
4            keywords_aggregate         8         0.750              0.0000
5                keywords_ratio         8         0.375              0.0000
Average correct rate: 0.58

python3 main.py \
  -db postgres \
  -q data/instruct_advanced_postgres.csv \
  -o results/openai_gpt3.5_turbo_advanced.csv \
  -g oa \
  -f prompts/prompt_openai.md \
  -m gpt-3.5-turbo-0125 \
  -c 0 \
  -p 20
Preparing questions...
Using all questions from data/instruct_advanced_postgres.csv
Correct so far: 30/64 (46.88%): 100%|███████████████████████████████████████████████████████████████████████████████████████████| 64/64 [00:08<00:00,  7.64it/s]
                 query_category  num_rows  mean_correct  mean_error_db_exec
0         instructions_cte_join        16        0.5625              0.1250
1       instructions_cte_window         8        0.1250              0.5000
2        instructions_date_join        16        0.3750              0.1875
3  instructions_string_matching         8        0.7500              0.0000
4            keywords_aggregate         8        0.6250              0.0000
5                keywords_ratio         8        0.3750              0.0000
Average correct rate: 0.47

Note that this isn't rigorously refined so there will be room for improving on these questions

[rishsriv]

Thanks for adding this!