0W standby circuit for switch?

HI, I am thinking about possibilities for an RF remote control circuit that only powers on when a button is pressed, thus would have 0W standby power.
Here is an idea that I have come up with.
The top lamp would be the Arduino, the left switch would be an output that the Arduino sets to LOW when it is done sending, to cut it's own power supply. Else it is set to high so the MOSFET gate stays high, so the RC part only has to hold the voltage until the Arduino can switch the pin to output. The gate of the MOSFET could instead also be an EN pin of an ultra-low dropout LDO, which I would use to get a LiPo down to 3.3V (for the RF module).
Due to the long time constant of the RC part a brownout shouldn't be an issue, and due to the low overall power draw, even if the MOSFET goes into saturation due to low gate voltage at some point (e.g. because the Arduino failed), that shouldn't be an issue.

What would be the startup time to the first line of setup() if I burnt the sketch without bootloader? Just the time set by the fuse?

You have a low-side switch enabled power to the load - is the load the Arduino? If so your
circuit cannot work as it puts negative voltage (w.r.t. Arduino ground pins) on output pin you
are using to latch the switch.

You need something like a CMOS flipflop permanently powered and positive-edge triggered on
both set and reset inputs, so it just sits in the same state when stuff is powered down, and you
need high-side switching so ground is common, everything is less confusing then! Flip-flop
drives logic-level p-FET as high side switch.

Yes, I can see that ... I thought about going for a flipflop before. The 74LVC1G74 could work, couldn't it? I cannot really see the power consumption in the datasheet. 100mA doesn't really make sense for a CMOS device ...

The goal is to built a keyfob for mysensors, including at ATSHA chip to make it usable for security related stuff like garage doors and alarm systems. If possible, it should run a long time on a coin cell. So everything should be as small as possible (well, no, I cannot solder that, but at least SMD) and really low current drawn when not in use.

In that course: Any suggestions for an ultra low drop-out regulator for 3.3V with an enable pin (until now, I always used an HT7333 without one)? The NRF2401L doesn't seem to like more, so I have to reduce the LiPo voltage.

Perhaps I should start a brain storming project guidance ..

Why not just build standalone arduino, use sleep to get power usage way the he'll down (see Nick gammon's page on avr power consumption), wake on button presses with pcints... And run it straight off your battery so you don't need the overhead of a regulator...

The regulator is for the NRF24L01, not the ATMEGA. That has a max voltage of 3.5V, AFAIR. If I onky run the RF chip over tje regulator, I also need a level shifter ...
Regarding sleep mode: Sure, could do that, but it is more than 0W.