I've seen datasheets for certain ICs recommend using a 100 kOhm (or so) pulldowns or pullups for setting configuration inputs. What could be a typical reason for this? Normally I would just tie the pin directly low or high.
The modern reason is a pull resistor lets you override it for debugging or change your mind and not populate it. A hard wired connection lets you do neither.
The historical reason was that old designs which run off BJTs instead of MOSFETs need pull resistors for current limiting since a direct connection to a power rail directly injects current into the diode-like base-emitter junction.
CORRECTION: Kevin White states:
" TTL ICs would recommend a pull-up resistor because the absolute maximum voltage for logic inputs was 5.5V whereas the maximum voltage for the power input pins was 7V. A surge on the supply rail to a voltage between 5.5v and 7v could damage the input transistor while the majority of the IC survived. It was considered good practice to provide current limiting to the inputs in the form of a resistor so that the device as a whole would survive a 7V surge."
This may not be what the OP is seeing but I’ve seen datasheets for some microprocessors that overlay configuration pins on pins for other input functions. The configuration is sampled only on reset, after which the pins assume their normal functions.
Using pull up/down resistors for configuration allows the configuration to be read but not interfere with the normal functioning once the chip is out of reset.
Probably the main reason a pull up or pull down resistor is required is because there isn't on on the device and leaving the input floating (open) means it would have an indeterminate logical state.
Consequently it's effect on the operation of the device is undefined.
An open input could also pick up stray signals and noise and it would appear to be switching from 0 to 1 at random; again with undefined consequences.