A "sufficient" shoot'em up it's not too complicated to make, you can get an OK-ish engine in a week, but then you have to add content (stages, enemies, different AIs, etc). IMHO hat's the main difficulty of the genre, because a lot of work is required for just a few minutes of gameplay.
I've made myself a couple of these games and they're fun to make, but it's not too rewarding (eg, I made this one in a month: http://www.usebox.net/jjm/flax/ -- and it's ~6 minutes to get to the end of the three stages).
BTW, I did a shmup for Android last year and decided to remove the spaceship altogether because I found it a little bit awkward to move it with a finger:
I did it in just 23 days by sticking to simple pixelated graphics for the aliens and only drawing UI and powerups. I also borrowed music with CreativeCommons license and only created the sfx myself. This gave me time to add various boss levels which is what makes the game interesting (and takes time as you wrote).
Also, powerups can be upgraded between games, so this got me a lot longer playing time than 6 minutes. You play to pass as many levels possible, but also to collect coins to upgrade your weapons which in turn enable you to pass more levels the next time and so on. There are 14 different bosses and you can hardly beat bosses 9+ without upgraded weapons.
There was a good ExtraCredits episode on YouTube explaining game loops (I can't find it now, but I recommend watching all EC episodes to any game developer). I didn't know about it at the time, but Drone Invaders fits great:
- smallest loop: kill enemies in a wave, reload guns
- bigger loop: every 10 waves there is a new boss level
- even bigger loop: when you die, upgrade your weapons and try to go even further
- biggest loop: reach the Fleet Admiral rank but hitting certain score, number of destroyed enemies, level reached and hiscore achieved over lifetime of playing.
I an monitoring the stats (the game sends anonymous usage info at the end of each play), and there are some players playing the game for over a month.
I was willing to use a WebView with Cordova, but Kit Kat introduced a bug (not hw acceleration in canvas 2D!) and I was getting 2 FPS instead of the expected 30.
Crosswalk works pretty well, although has the inconvenient of being a native component.
Not everyone upgrades their phone every year or two. I know plenty of people who still have an Android 2.x smartphone because they have no need to upgrade.
(That said, I don't think many of them would play this game.)
Some people are using devices that are hard to upgrade and are stuck with older versions. According to Google Play stats, I have about 11% of players using Android 2.3 for my game.
I use 2.3.3 as the lowest supported version, since most important features were introduced with 2.3.3. At least, I don't know any reason to require 4.0 to be minimum, so why lose 11% of your users. I was curious if OP used some important feature that would require 4.0.
To be honest, if you are indie developer and you don't have very aggressive IAP like Clash of Clans or Throne Rush, there is no money from people that buy apps. I've seen cool indie games like TripleTown having trouble monetizing. I bet one could make more money from Ads (although I hate ads, I hate them less than whale-catching-IAPs).
I've made myself a couple of these games and they're fun to make, but it's not too rewarding (eg, I made this one in a month: http://www.usebox.net/jjm/flax/ -- and it's ~6 minutes to get to the end of the three stages).