Rige
You can make two flows within survey flow with same blocks but make the second one on a branch logic of dummy embedded data e.g. response=manual and no randomizer and first one the opposite of it eg. response ≠ manual. Then every time you want to fill for all just query this parameter in URL the second branch will execute with all blocks.
Hope it helps!

Thank you Deepak for your comment.
Just to see if I understood right; if I build a second flow, I’m able to answer all the questions, even in blocks, where number of questions are limited with randomization (to show f ex 2 questions out of 5)? I think new flow without url parameters can’t overide limited question randomization in block? I do understad, that I can use url parameter to show all blocks but how to show all questions inside block when there is radomization within block?

https://community.qualtrics.com/XMcommunity/discussion/comment/55615#Comment_55615Well randomisation within block would be challenging I presumed you had it in survey flow just.
You won't be able to override that.

Rige,
You may want to consider putting the manual responses in a spreadsheet, one per row, then import the responses as a csv.
An alternative to Deepak's solution that could work for block and questions, is to set up embedded data flags for the various blocks and questions using randomizers. The number of flags set would be based on the whether it is a manual or regular response. Then for blocks, put your blocks under branches based on embedded data flags that are inside a randomizer but display all the items (only the true branch condition blocks will display). For questions, add display logic based on the embedded data flags to all the questions. Randomize the questions in the block, but don't limit the number of questions displayed (only the true display logic question will display).

Thanks TomG!
I will try the solution you suggested. Let you know if succeeded. However, csv is always option. Actually I already made a copy of survey and cleared all the randomizations off the copy one. Then used that copy for manual responses and expoted csv out and imported csv to original. Emptied the copy to handle upcoming new manual responses and ”play around again”. However, would like to skip all that hussle.:)