fix(testset): for generating testset with new docs

[Manav916]

fixes #848

This pull request introduces updates to the TestsetGenerator and ComplexEvolution classes within the generator.py and evolutions.py files respectively. The primary focus of these changes is to improve the handling of new documents during test dataset generation, ensuring that data from previous documents does not interfere with the generation process.

Before

Here row 2 and 4 are generated from the document used for the previous generation

After

Here all rows generated are based on the new document provided

[AliHaider0343]

how we could use it i am having the exact same issue how we will be able to get the same results?

[XinTaoIntel]

When this will merged? When are having the same issue. After this is merged, any code change I need to make?
Currently I'm just use standard generation method without using evolution. (Do we need to add evolution part to not have this issue?)
testset = generator.generate_with_langchain_docs(
documents,
test_size=3,
distributions={simple: 0.5, reasoning: 0.25, multi_context: 0.25},
)