If the use of Bitcoin increase in the future, is the 7 transactions per second going to be a bottleneck?
Assuming Bitcoin continues to grow in usage and we run out of possible soft forks that fit more transactions into a block without needing a hard fork (e.g. SegWit, Taproot) then yes fees will continue to rise if the block weight limitation remains the same.
And if so, can the block size be easily increased to grow network capacity?
I don't have a crystal ball but I suspect it won't be easy although hopefully not quite as fraught as the block size wars of 2015-17. There has to be network consensus over a block weight increase which will be extremely difficult. In addition if it is indeed a hard fork it will need years of planning and preparation.
Will this be a hardfork just like Bitcoin Cash?
If there are no alternative soft fork proposals which offer scaling potential then I suspect a hard fork will be proposed (though no guarantees on whether it obtains consensus) under the assumptions I laid out above. Bitcoin Cash's initial hard fork was a 8x increase (it later increased the block size limit to 32MB) and any proposed hard fork for Bitcoin would need to be a smaller multiple than that to have any chance of obtaining network consensus.
And why didn't Satoshi choose a block size greater than 1MB to avoid this potential bottleneck?
As far as I know Satoshi introduced the block size limit to address DoS (denial of service) issues. At the time 1MB seemed reasonable and still does to some respects. Even in 2021 some people think it should be higher, some people think it should be lower which suggests that many years ago it was about right.
Thanks to Murch for the suggested edits.