Backport PR #60324: REF: centralize pyarrow Table to pandas conversions and types_mapper handling #20437

Job	Run time
Linux-32-bit	20m 11s
Linux-Musl	17m 55s
python-dev (macOS-13)	33m 28s
python-dev (macOS-14)	14m 36s
python-dev (ubuntu-22.04)	23m 59s
python-dev (windows-latest)	26m 35s
macos-13 actions-311.yaml	43m 14s
macos-13 actions-312.yaml	43m 22s
macos-13 actions-310.yaml	39m 13s
macos-13 actions-39.yaml	1h 1m 14s
macos-14 actions-310.yaml	34m 18s
macos-14 actions-311.yaml	31m 18s
macos-14 actions-312.yaml	27m 6s
macos-14 actions-39.yaml	37m 25s
windows-latest actions-312.yaml	47m 22s
windows-latest actions-310.yaml	39m 40s
windows-latest actions-311.yaml	42m 3s
windows-latest actions-39.yaml	35m 54s
ubuntu-latest actions-310.yaml	52m 59s
Copy-on-Write 3.10 (warnings)	27m 54s
Copy-on-Write 3.10	28m 4s
ubuntu-latest actions-311.yaml	49m 5s
Future infer strings (without pyarrow)	47m 32s
Locale: zh_CN	27m 30s
Copy-on-Write 3.11 (warnings)	26m 30s
Copy-on-Write 3.11	26m 44s
Locale: it_IT	26m 58s
Downstream Compat	5m 52s
Numpy Dev	23m 19s
Pyarrow Nightly	23m 40s
ubuntu-latest actions-312.yaml	37m 56s
Future infer strings	39m 36s
Copy-on-Write 3.12	30m 21s
ubuntu-latest actions-39.yaml	50m 22s
Copy-on-Write 3.9 (warnings)	28m 15s
Copy-on-Write 3.9	28m 13s
Minimum Versions	27m 53s
Pypy	7m 39s
	20h 35m 15s

Provide feedback