read_le<T>/read_be<T> incorrectly evaluate float

[joekale]

While using read_be with ibeo_lux driver I noticed that scan_point xyz values were abnormally large integers. Ran a small test to recreate with read_le:

1. Entered 123.4567 as value in IEEE floating point converter: https://www.h-schmidt.net/FloatConverter/IEEE754.html
2. Copied Hexadecimal representation (0x42f6e9d5) from converter and generated an int using a calculator (1123477973)
3. wrote small commandline program to take integer as input and run through read_le

output as follows (bytes print in little_endian order):

./network_util_test 1123477973
input: 1123477973
bytes: 213, 233, 246, 66
Hex Representation: 0xd5e9f642
Value of encoded float: 1123478016.000000