[hw/sw] update flash load linker script, data_interleaved and data_flash_only sections

[davideschiavone]

• flash_load linker now is aligned with the on_chip linker, with different regions for code and data

• added data_interleaved section for mapping data on those memory banks

• added flash_only section for big data that cannot be mapped to sram

• modified crt0 to use the flash BSP

• bump to 2KB the pre-fetched code that the bootROM reads in flash_load mode as more functions are used in crt0 due to BSP

• test all APPs on-chip

• test all APPs flash_load

• test all APPs flash_exec

• check spi_write examples on FPGA

• all tests work except for the flash_only data that reports all 0s value read back.

• check spi_write examples on sim

• check spi_read examples on FPGA

• check spi_read examples on sim

• add flash_data_only example

• add data_interleaved example

• check performance differences on 1 app (e.g. matadd)

• check freeRTOS on FPGA

[davideschiavone]

replacing #398

[davideschiavone]

note to myself @davideschiavone,

So far the SPI examples fail (when flash_load is used) as they have been built under the assumption that LMA=VMA,
whereas data in FLASH are not anymore at the same offset

We may need to use symbols from the linker script to get the offset displacement and add it to the application data addresses
to do the flash2memory copies

https://stackoverflow.com/questions/8398755/access-symbols-defined-in-the-linker-script-by-application

https://downloads.ti.com/docs/esd/SPRUI03/using-linker-symbols-in-c-c-applications-slau1318080.html