perf: use switch to detect method

[gurgunday]

As more methods are added, the difference should get more significant

Also switch is more readable in my opinion

[gurgunday]

The benchmark label seems broken

[github-actions]

Node: 16
PR: [1] 1746k requests in 30.04s, 328 MB read
MAIN: [1] 1758k requests in 30.05s, 330 MB read

---

Node: 18
PR: [1] 1537k requests in 30.04s, 289 MB read
MAIN: [1] 1501k requests in 30.04s, 282 MB read

---

Node: 20
PR: [1] 1516k requests in 30.05s, 285 MB read
MAIN: [1] 1450k requests in 30.05s, 272 MB read

[github-actions]

Node: 16
PR: [1] 1570k requests in 30.04s, 295 MB read
MAIN: [1] 1568k requests in 30.04s, 295 MB read

---

Node: 18
PR: [1] 1360k requests in 30.04s, 256 MB read
MAIN: [1] 1340k requests in 30.04s, 252 MB read

---

Node: 20
PR: [1] 1481k requests in 30.05s, 278 MB read
MAIN: [1] 1441k requests in 30.04s, 271 MB read

[gurgunday]

Second one is slightly more relevant as it tests contentType parsers and mainly uses 'POST' instead of 'GET'

GET gets matched immediately in both cases so it's not very useful

[github-actions]

Node: 16
PR: [1] 1768k requests in 30.04s, 332 MB read
MAIN: [1] 1785k requests in 30.04s, 335 MB read

---

Node: 18
PR: [1] 1526k requests in 30.04s, 287 MB read
MAIN: [1] 1522k requests in 30.04s, 286 MB read

---

Node: 20
PR: [1] 1559k requests in 30.04s, 293 MB read
MAIN: [1] 1561k requests in 30.04s, 293 MB read

[github-actions]

Node: 16
PR: [1] 1442k requests in 30.04s, 271 MB read
MAIN: [1] 1491k requests in 30.04s, 280 MB read

---

Node: 18
PR: [1] 1377k requests in 30.04s, 259 MB read
MAIN: [1] 1356k requests in 30.04s, 255 MB read

---

Node: 20
PR: [1] 1332k requests in 30.05s, 250 MB read
MAIN: [1] 1274k requests in 30.04s, 239 MB read

[gurgunday]

This won't be needed after #5419