If it really is 1095 it will not truly air harden. True hardening involves entrapment of carbon in solution to form martensitic structure, and you need to cool from 1430-1500F to below 950F in less than .5 seconds in order to achieve that in 1095. But what can happen is carbide segregation into course networks which will be the equivalent of sheets of diamonds mixed into a ball of soft clay. A file will simply skate off from this but a Rockwell penetrator will bury itself into it. I implore you- no not use lamellar annealing practices with 1095, it will only make this problem much worse. Lamellar annealing is slow cooling from critical, ala wood ashes, vermiculite, or cooling forge, will only make this segregation problem much, much, worse. There are two methods of annealing, lamellar and spheroidizing, the latter is a subcritical or isothermal method that you want to use for 1095. Normalize exactly as you have done and then follow it up with several heats to around 1300F (reddish color) but never allow the magnet to stop sticking, if you do, start over again. This subcritical cycling will take all of those carbides sheets and turn them into very tiny spheroidal globules within the iron matrix for a very soft structure. If you find that you still have some issues with hard areas then the carbides are not spheroidizing evenly, to help with this you can follow up the normalization with a heat to 1475F and a quench to trap things into solution. When this solid solution is then spheroidized at around 1300F the carbides will be much more finely and evenly distributed.