Each I2S interface can capture up to 16 channels, but if you need to interface multiple microphones and they cannot be connected to the same I2S, then yes you may need to use more than one interface.
Is there a Linux driver available for this MEMS mic? Or does it work with the dummy spdif driver (default codec driver) for the I2S interfaces? If it works like the Adafruit MEMS mic  which does not require a specific codec driver, then it could be possible to configure one I2S as the master and one as a slave and then use the BCLK and FSYNC from I2S1 (40-pin header) to drive I2S2 (30-pin header) as well as the MEMS mics.
Please note that the I2S signals on the 40-pin header can operate at 1.8V or 3.3V and this is configured by jumper J24. The I2S signals on the 30-pin header only support 1.8V and so you will need to configure J24 so that the I2S on the 40-pin header are 1.8V. Placing the jumper in position 2-3 should configure for 1.8V operation.
With regard to pinmux, you can check the current configuration by …
$ sudo grep "dap1\|dap2" /sys/kernel/debug/tegra_pinctrl_reg
For I2S2/DAP2 you want something like …
$ grep dap2 Linux_for_Tegra/bootloader/t186ref/BCT/tegra186-mb1-bct-pinmux-quill-p3310-1000-c03.cfg
pinmux.0x02434018 = 0x00000440; # dap2_sclk_pc1: i2s2, tristate-disable, input-enable, lpdr-disable
pinmux.0x02434008 = 0x00000400; # dap2_dout_pc2: i2s2, tristate-disable, input-disable, lpdr-disable
pinmux.0x02434000 = 0x00000458; # dap2_din_pc3: i2s2, pull-up, tristate-enable, input-enable, lpdr-disable
pinmux.0x02434010 = 0x00000440; # dap2_fs_pc4: i2s2, tristate-disable, input-enable, lpdr-disable
The above should allow the I2S2 to function as either slave or master.
Finally, it should be noted that Tegra has an Audio Multiplexer (AMX)  that can be used for multiplexing multiple audio streams into a single stream of up to 16 channels. So you could route the output from the I2S interfaces to the AMX and then capture the audio from both I2S interfaces as a single audio stream. Assuming that there are 16 of less channels in total.