Exception thrown: LOSSLESS + OpenMP

[ShatrovOA]

Dear SZ3 Developers,

Been checking out SZ3 lately. Noticed that it also supports lossless compression via ZSTD.
However it breaks when OpenMP is enabled:

terminate called after throwing an instance of 'std::length_error'
  what():  The buffer for compressed data is not large enough.

Tested using sz3 executable with testfloat_8_8_128.dat data.
Number of cores do not matter
Changing bytesCap to a bigger value at https://github.com/szcompressor/SZ3/blob/master/tools/sz3/sz3.cpp#L133 also do not matter

Turning off OpenMP makes everything work.

Configuration used

# SZ3 Configuration File
#
# This file configures SZ3 compression and decompression settings.
# Uncomment and modify the settings below as needed.
# For detailed explanations of each option and their valid values, refer to SZ3/utils/Config.hpp.
#
# Default values are used if settings are commented out or omitted.

[GlobalSettings]

# Compression algorithm
# Options: ALGO_LORENZO_REG, ALGO_INTERP_LORENZO, ALGO_INTERP, ALGO_NOPRED, ALGO_LOSSLESS, ALGO_BIOMD, ALGO_BIOMDXTC
CmprAlgo = ALGO_LOSSLESS

# Error bound mode
# Options: ABS, REL, PSNR, NORM, ABS_AND_REL, ABS_OR_REL
ErrorBoundMode = REL

# Absolute error bound (used with ABS mode)
AbsErrorBound = 1e-3

# Relative error bound (used with REL mode)
# RelErrorBound = 1e-3

# PSNR error bound (used with PSNR mode)
# PSNRErrorBound = 70

# L2 norm error bound (used with NORM mode)
# L2NormErrorBound = 0.333

# Enable OpenMP for parallel processing (YES or NO)
OpenMP = YES



[AlgoSettings]

# Advanced algorithm settings for tuning (not recommended for regular users)
# Uncomment only if you understand the implications.

# InterpolationAlgo = INTERP_ALGO_CUBIC
# InterpolationDirection = 0
# Lorenzo = Yes
# Lorenzo2ndOrder = No
# Regression = Yes
# Regression2ndOrder = No
# BlockSize = 6
# QuantizationBinTotal = 65536