Improve saving-data guide

[janbuchar]

https://crawlee.dev/python/docs/introduction/saving-data#using-a-context-helper should put emphasis on using the push_data helper, Dataset.open().push_data() should only be mentioned later in the article

[vdusek]

I went through the Introduction guide again, and I disagree with the proposed change. This approach aligns with how we explain concepts elsewhere: first, we delve into what's happening under the hood, and then we introduce shortcuts/helpers. For example, check the First crawler page - initially, we explain the RequestQueue and how to manually append requests to it. Only afterward do we show that you can simply pass requests directly to crawler.run(requests). Also note, that this is just taken over/rewritten from the TS version - not from my head, but I like this approach.

[janbuchar]

Thing is, in JS crawlee we try to push users towards using the context helpers and avoiding directly touching the Dataset class, for instance. The reason is consistency (there are way to many ways to work with storages and we should have a preferred option) and adaptive crawling where we "override" the helpers so that we can inspect and compare the results of multiple executions of a request handler. The adaptive crawler even enforces this by disabling the non-context ways of accessing the storage. I'm open to any thoughts on this though.