Provide lux value as a float rather than uint32_t

[amaschas]

One of the advantages VEML6030 is its extreme sensitivity, which allows for a resolution well under 1 lux. The current implementation here, however, casts the light reading as an integer rather than a float, which limits the effective resolution of the sensor when using this library. Changing the return value of functions involved in reading and calculating the lux value of the sensor reading to float should rectify this. Alternatively, an additional function could be added to return the value as a float, while keeping the original implementation intact.