Avoid copying elements when parsing

[MichaReiser]

This PR tries to reduce the number of elements that we copy by

• Removing the need for calls to vec.extend
• Removing the need for vec.insert(0, element)
• Use lalrpop named arguments instead of map|x| x.1) to ignore some tokens

Performance

I hoped that this would improve performance more but it remains mostly unchanged

group                        base                                   opt
-----                        ----                                   ---
parser/large/dataset.py      1.00      3.5±0.01ms    11.7 MB/sec    1.00      3.5±0.05ms    11.7 MB/sec
parser/numpy/ctypeslib.py    1.02    664.1±1.66µs    25.1 MB/sec    1.00    654.0±1.27µs    25.5 MB/sec
parser/numpy/globals.py      1.02     67.1±0.56µs    44.0 MB/sec    1.00     65.5±0.44µs    45.1 MB/sec
parser/pydantic/types.py     1.00  1416.3±19.83µs    18.0 MB/sec    1.01   1430.9±4.21µs    17.8 MB/sec