Thank you for raising issue with this description and persistence to get it clarified. The "ESP32 Hardware Design Guidelines" will be updated.
C5 (10 nF) that connects to CAP1 is required to suppress the noise of internal voltage reference and cannot be be removed. RC circuit between CAP1 and CAP2 pins may be omitted under certain conditions. This circuit is used when entering Deep-sleep mode. During this process, to minimize power consumption, the voltage to power ESP32 internals is dropped from 1.1 V to around 0.7 V. The RC circuit is used to minimize the period of the voltage drop. If removed, this process will take longer and the power consumption in Deep-sleep will be higher. If particular application of ESP32 is not using Deep-sleep mode, or power consumption is less critical, then this circuit is not required.