At a few MHz the cable may not be so critical. But without a low-pass filter, the higher frequency components will still show up on the scope and might confuse you.
The divider will be at the probe to keep the reactance in parallel with the divider down. Any reactance (capacitance or inductance) after the divider gets divided by the division ratio. So if the cable has say 60 pF of capacitance, and the divider is behind the cable, the probe has an input capacitance of at least 60 pF. While if there is a 10x divider in front of the 60 pF cable, then the contribution to the input capacitance is only about 6 pF. See figures 2-15B and 2-16A of the Tek document which show the 9 MOhm series resistor with cap in parallel at the probe end.
Figure 2-4 shows a very rough schematic of what a 100x probe might look like, with a 9.9 MOhm resistor in series and a 111 kOhm resistor in parallel with the 1 MOhm scope input to form 100 kOhm, a /100 divider. The advantage of this that even with the probe unplugged, the voltage at the BNC connector is still roughly 1/100th of the input voltage, as opposed of the alternative design of a 99 MOhm series resistor which would present a much higher voltage when the probe is disconnected, or the scope is set to AC coupling.