Because it takes time to compute the statistics. More the less power your device had. Imagine a fairly active network with a dozen clients and Pi-hole being hosted on a Raspberry Pi Zero. Computing even the last 7 days might be something which needs several seconds. And it is something the database computes internally, so it is also not easily interruptible.
We do not trigger a processing when the page is opened to avoid an additional delay until the user can request what they want.
The long-term statistics are meant as a research tool, not something you use several times every day. It is more a scalpel rather than a kitchen knife. IMO it should allow the user to get what they want without being in their way. Hence, we designed it the way it is.