But I'm not able to confirm through research that it is correct to say the first stage or second stage is the first readable sector
The first sector is the first sector. As far as the legacy BIOS was concerned, it would only load the first sector of a drive into memory and pass control to whatever is in there (this "stage 1"). The BIOS itself is never aware of, nor does it care about, any "stage 1.5", "stage 2", etc. (which themselves are specific to GRUB — while other bootloaders typically apply similar methods, there is no rule requiring so). In fact, the BIOS doesn't even care about what "stage 1" is or does, as long as that sector can be read from the disk (though some firmware does look for a flag marking it as bootable).
When people say "the first readable sector" are they talking from the operating system or firmware (BIOS/UEFI) perspective?
Generally the same thing, though there are exceptions. "first readable sector" in itself doesn't tell you anything about whose perspective you are reading from. This would depend on context.
Does a difference exist between what the operating system can see of a mbr disk in comparison to what the computers fimware can see (UEFI/bios)?
Generally, no — assuming the OS has direct access to the physical disk. One exception is if virtualisation is involved.
For GPT Disks does the BIOS Boot partition only exist for backwards compatibility with BIOS?
Yes. Most GPT-based installations do not have a BIOS Boot Partition at all. Instead, they only have a EFI System Partition designed for boot from EFI-compatible firmware.
There are exceptions to the above. For example, the Host Protected Area can effectively hide part of the drive, however, the HPA is at the end of addressable space, not the beginning.
Other abstractions can come into play. For example, some types of RAID work at the operating system driver level, leaving the motherboard firmware and the operating system with different views of the drive. Other types will load an option ROM into the motherboard firmware, which can result in the firmware seeing a different "first sector" from what is exposed by the disk. Intel's "fake" RAID is an example of this.
And even what the disk exposes as the "first sector" (first block, with LBA) can be different from what is physically on the disk. We are long past the old days when the disk controller was separate from the physical disk, and these days we no longer know exactly how the disk's internal structure is laid out. If you really want to get technical, you could claim the "first readable sector" differs from what is internal to the drive, but this is largely pointless. Modern drives can also remap "unreadable" (damaged) sectors to spare areas internally — this is transparent to everything external.
Side note, first readable sector actually appears in a comment in the Linux 2.4 kernel source (removed in 2.6) for the Advanced Disc Filing System in acorn.c. Note that this file system (and the system it was introduced with) is from the 1980s and long disused. From Wikipedia:
Hard disc support in ADFS used the same format as L format floppies in terms of 256-byte blocks;[5] only the underlying arrangement of tracks and sectors differed depending on the actual drive used, but this was managed by the SCSI controller. It interfaced to a ST506/ST412-based Winchester unit via the BBC Micro's 1 MHz Bus, an Acorn-designed interface card (1 MHz Bus to SCSI adapter) and an off-the-shelf Adaptec SCSI controller (SCSI to ST-506 adapter).
