Simply, because the Elizabeth line does not have the capacity to absorb GWR’s passengers as well as its own; and Elizabeth line doesn’t have the resources to run additional services to cope with the demand that would ensue.
A weekend service to/from the West of England can have easily north of 900 passengers onboard at Paddington when there’s zero disruption.
***
There isn’t a palatable, quick solution to this.
Services at weekends are already bursting at the seams with overcrowding.
There aren’t the resources to run anywhere near the planned weekend service; let alone any extras to relieve that overcrowding.
Any exercise to bring Sunday into the working week properly will involve recruiting and training a significant headcount increase - and thus be years down the line.
With the level of overcrowding as it is, you can’t then reduce the service further to match crew levels.
You could (theoretically) make traincrew diagrams more efficient; but that efficiency comes at the expense of the flexibility that is used on the day to recover from disruption. Some would even argue we’ve already gone too far in this regard and we need a return to less efficient but more resilient diagramming to reflect the typical level of disruption encountered as opposed to planning a perfectly ontime railway