[Benchmarks] Make partitions default to number of cores instead of 2

[andygrove]

Which issue does this PR close?

Related to discussion in #7001

Rationale for this change

By default, the benchmarks have been running with two partitions (i.e., concurrency of two), which is unrealistic given the number of cores on modern laptops. This PR defaults to num cores instead. We can still override this and explicitly set the number of partitions.

This is the difference this makes for me (Threadripper with 24 cores).

┏━━━━━━━━━━━━━━┳━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query        ┃      main ┃ bench-default-partitions ┃        Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1     │  655.23ms │                 283.37ms │ +2.31x faster │
│ QQuery 2     │  148.53ms │                 160.08ms │  1.08x slower │
│ QQuery 3     │  277.50ms │                 215.95ms │ +1.28x faster │
│ QQuery 4     │  166.20ms │                 131.36ms │ +1.27x faster │
│ QQuery 5     │  334.64ms │                 248.19ms │ +1.35x faster │
│ QQuery 6     │  144.79ms │                  97.53ms │ +1.48x faster │
│ QQuery 7     │  718.12ms │                 367.63ms │ +1.95x faster │
│ QQuery 8     │  385.01ms │                 293.77ms │ +1.31x faster │
│ QQuery 9     │  571.56ms │                 387.59ms │ +1.47x faster │
│ QQuery 10    │  459.09ms │                 388.16ms │ +1.18x faster │
│ QQuery 11    │  143.07ms │                 119.23ms │ +1.20x faster │
│ QQuery 12    │  257.05ms │                 244.25ms │     no change │
│ QQuery 13    │  447.24ms │                 385.33ms │ +1.16x faster │
│ QQuery 14    │  233.90ms │                 126.45ms │ +1.85x faster │
│ QQuery 15    │  192.98ms │                 143.86ms │ +1.34x faster │
│ QQuery 16    │  163.44ms │                 176.92ms │  1.08x slower │
│ QQuery 17    │  335.34ms │                 230.92ms │ +1.45x faster │
│ QQuery 18    │ 1188.63ms │                 678.13ms │ +1.75x faster │
│ QQuery 19    │  360.65ms │                 193.74ms │ +1.86x faster │
│ QQuery 20    │  356.26ms │                 168.59ms │ +2.11x faster │
│ QQuery 21    │  825.86ms │                 491.50ms │ +1.68x faster │
│ QQuery 22    │  163.89ms │                 201.47ms │  1.23x slower │
└──────────────┴───────────┴──────────────────────────┴───────────────┘

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?

No

[alamb]

We are going to need some bigger benchmarks :) -- by the time the queries are in the 250ms range, there is a lot of noise with whatever else is going on in the system :)

[alamb]

Thanks @andygrove

[alamb]

I think SessionConfig already defaults to the number of CPUs, so we can probably just avoid setting this at all if self.common.partitions.is_none()