MIPS: Support for relocations that pair multiple R_MIPS_LO16 against a single R_MIPS_HI16

[bmourit]

According to https://www.linux-mips.org/pub/linux/mips/doc/ABI/mipsabi.pdf, a R_MIPS_LO16 without a R_MIPS_HI16 (orphaned) should be matched with the previous R_MIPS_HI16. This isn't currently taken into consideration in mips relocation.

This isn't a bug, but rather an enhancement. I think a simple variable to hold the previous value to pair with an orphaned lo16 would be sufficient here. If I get time I'll post a pull request for this.

[PeterMatula]

Thanks.
Do you have some test case that we could add to regression tests? Did you find this problem in a sample you could share - you don't mind if we add it to a public test suite.