Prompt engg to get better data.

[rachitavya]

Current data corresponding to this latest PDF can be checked here.

We're still loosing few of the crops like paddy and pulses.

cc: @aashutosh-samagra

[rachitavya]

This will require:

• [x] Improving prompts
• [x] Including summarization in the pipeline
• [x] Improving fallback #6
• [x] Preventing llm from creating headings itself out of content
• [x] Cleaning step to remove generic suffixes like cultivation in front of crops. (e.g. wheat cultivation)

[rachitavya]

PR merged.