[PWGLF] Improvements in data model for strangeness, strangeness tof pid cleanup

[ddobrigk]

This PR:

• Saves approximately 10% disk space in derived data by:
  • detaching TOF information from daughter track information to save disk space and links the new table with an index
  • compresses TPC Nsigma into tiny format (from -6.35 to +6.35 in 0.05-sized increments) while keeping the interface unchanged for a smooth transition
• Provides converters to keep existing data usable
• Cleans up the strangenesstofpid further and also alters the second processing algorithm to use adjusted primary TOF information as opposed to a standalone calculation, effectively leading to e-loss-level performance as reported in the PAG but with negligible CPU overhead. The adjustment versus primary TOF Nsigmas has been tested to lead to an Nsigma distribution centered around zero, which would not be the case if the primary-particle Nsigmas were used. More testing to take place in continuation.

Tagging @romainschotter @ercolessi @gianniliveraro @njacazio

[github-actions]

O2 linter results: ❌ 296 errors, ⚠️ 0 warnings, 🔕 0 disabled