Generating term vectors on the fly

Suppose, I have more than 1mil documents where I have a text field lets say "Contents". We have not enabled termvector for the index.

Now when we want to find count of occurrence of a word lets say "data" in "Contents" , I am thinking of generating term vectors on the fly. But had below queries.

Can you please tell me what do we mean when we say " Note that for the field text , the terms are not re-generated." ? - refer below image.

Additionally what will be impact on performance when we use term vector on the fly for text field? Do you have any data which you can share?