read_bits_int_{be,le}(): speed up using int.from_bytes()

[`int.from_bytes()`](https://docs.python.org/3/library/stdtypes.html#int.from_bytes)
was added in Python 3.2. Unsurprisingly, it is faster than our pure
Python loop in `read_bits_int_be()`:

```console
$ python3 --version
Python 3.14.2
$ python3 -m timeit -s 'buf = b"\xb5\x6c\x23"' $'res = 0\nfor byte in buf:\n    res = res << 8 | byte'
2000000 loops, best of 5: 185 nsec per loop
$ python3 -m timeit -s 'buf = b"\xb5\x6c\x23"' 'res = int.from_bytes(buf, "big")'
2000000 loops, best of 5: 127 nsec per loop
```

Our previous loop in `read_bits_int_le()` was particularly slow compared
to `int.from_bytes()`:

```console
$ python3 -m timeit -s 'buf = b"\xb5\x6c\x23"' $'res = 0\nfor i, byte in enumerate(buf):\n    res |= byte << (i * 8)'
1000000 loops, best of 5: 339 nsec per loop
$ python3 -m timeit -s 'buf = b"\xb5\x6c\x23"' 'res = int.from_bytes(buf, "little")'
2000000 loops, best of 5: 127 nsec per loop
```

Author: generalmimon

kaitaistruct.py

@@ -323,8 +323,7 @@ def read_bits_int_be(self, n):
             # 9 bits => 2 bytes
             bytes_needed = ((bits_needed - 1) // 8) + 1  # `ceil(bits_needed / 8)`
             buf = self._read_bytes_not_aligned(bytes_needed)
-            for byte in buf:
-                res = res << 8 | byte
+            res = int.from_bytes(buf, "big")
 
             new_bits = res
             res = res >> self.bits_left | self.bits << bits_needed
@@ -365,8 +364,7 @@ def read_bits_int_le(self, n):
             # 9 bits => 2 bytes
             bytes_needed = ((bits_needed - 1) // 8) + 1  # `ceil(bits_needed / 8)`
             buf = self._read_bytes_not_aligned(bytes_needed)
-            for i, byte in enumerate(buf):
-                res |= byte << (i * 8)
+            res = int.from_bytes(buf, "little")
 
             new_bits = res >> bits_needed
             res = res << self.bits_left | self.bits