Add test-time scaling strategies like resampling and ensembling

[liana313]

Problem Statement

When running operations like sem_filter or sem_map, I would like to be able to specify a n_sample parameter and ensembling strategy

Proposed Solution

Allow users to specify a test-time-scaling strategy and ensembling strategy per operator. For example

df.sem_filter("the {abstract} is relevant to vector databases", n_sample=3, ensemble='majority_vote', temperature=1.0)

Use Cases

When I perform a sem_filter (eg "is this row relevant to XX"), the results often change for repeated trials, so it would help to have built in functionally

Checklist

• [ x] I have searched existing issues to avoid duplicates
• [ x] I have provided a clear problem statement
• [ x] I have considered alternative solutions
• [ x] I have assessed the impact and priority
• [ x] I am willing to contribute to implementation (if applicable)