wireless moisture data transmission

Below is what I would do if I were you.

I would completely skip the Pi (unnecessary design complications and need for additional hardware + code) and ensure the nodes send data directly to the AWS over wifi(for indoors) using MQTT.

For ensuring a longer battery life, I would to the following:

1. Ensure the nodes are in deep-sleep most of the times and wake up only when needed to send the sensor data and then go back to sleep. Something like below.

deep sleep-->timer elapse INT-->wakeup-->turn on radio-->connect to wifi-->connect to server-->read sensor data-->transmit data-->disconnect from server--> disconnect from wifi-->turn off radio-->reset timer/clear INT-->go back to deep sleep.

2. Your battery life will also depend on the rate at which you transmit the data. I would generally do it once in a few hours in your case.

3. To get a good estimate of how much power you node draws and how long it can last on battery. Measure the current drawn when a node is in deep sleep + measure the current drawn when full active and then do the math based on (2) and you can get a rough idea of how long the device could last on battery.

For suggestion for the micros, I would go with an AVR (something like the atmega1280 or 1281) and use an external module for wifi in case I decide to switch to a different one at some point in time, it would be easier.

You got the right idea, let's jump into your questions:

First of all, you need to think about the microcontroller itself. Since you said that you need wifi/bluetooth, I could recommend an ESP8266 (wifi only) or an ESP32 (wifi and bluetooth, also more expensive).

Now for the battery: there are some dev boards which already include all of the battery management stuff, like some LILYGO boards that can be easily found on AliExpress or the likes, with support for the commonly found 18650 cells. Another option would be to get a regular dev board with no battery slot, and add it yourself if you want something more involved.