there was a big discussion on how to solve this issue. A fork with bigger blocks was made (bitcoin cash) but it didn't gain traction.
This is misleading because it sounds like two routes were taken to scale "bitcoin" and one of them was "bcash" whereas in reality many people have been creating "copies of bitcoin" pretending they are one of those "routes" to solve issues bitcoin is facing, one of those shitcopies were bcash.
It also didn't gain any traction because it was an altcoin created by copying bitcoin's source code and blockchain like about 2 dozen others at the time (bgold, bsliver, btx, ...).
Any "bitcoin" fork needs to be voted in by the miners and if it gains supermajority, it can be categorized as one of those "routes" to take.
Basically, transaction data was still limited to 1Mb, but witness data (usually the signature) was moved to the end of the block,
Wrong.
Witness, similar to any other part of the tx (like version, output script, locktime, etc.) is part of the transaction and is included in the transaction itself not the "end of the block".
Basically this structure:
[version][input_count][outpoint][signature_script][output_count][amount][output_script][locktime]
turned into this:
[version][flag][input_count][outpoint][signature_script][output_count][amount][output_script][witness][locktime]
and no longer counted to this 1Mb limit (there is a limit for witness data aswell, but afaik, it's very seldomly reached due to the existing limit of 1Mb for the actual transaction data).
The 1 MB limit is no longer part of the checks performed in the consensus rules. We are now computing "weight" and the limit on that weight is 4 MB. The way weight is calculated means the stripped block will have <= 1 MB size.
the size of the transaction (excluding the witness data) basically determines how much transactions fit into a block...
Weight of the transaction...