it looks like the number of observation that fit your date range + taxon + life stage criteria is less than 10,000. so you have many options for how to proceed.
downloading is basically a 2-step process. you need to get a list of the photos to download*, and then you need to download them. assuming you have a list of photos already from the iNat CSV export, iNat API, or some other source, you can download them using your favorite scripting language or similar tool.
the way i would probably do it is by creating some cURL statements to do the download (since i would probably want to keep a record of exactly what i downloaded). here’s a post that talks about this process a bit: https://forum.inaturalist.org/t/best-way-to-download-600-images/37403/4.
you could probably automate the whole process from the command line using a combination of cURL + jq, or you could probably do it in R or Python, etc. i don’t have any such scripts to share though, although i’m sure you could look through GitHub and find something, if needed.
if the images you get are licensed (as opposed to all rights reserved), the image files should come from the AWS Open Data set, in which case there are few limits on how you need to download the files. if they include all rights reserved images for some reason, iNat will impose certain restrictions when downloading from their media servers. (there should be references to those in the post linked above.)
* it may be worth noting that the AWS Open Data set was designed probably to serve specifically as a potential training set, but it’s probably harder to use the AWS set metadata as a starting point for getting your list of image files to download in this case because you’re not dealing with a lot of observations and because you have some more niche filter criteria — specifically, life stage.